Mike HeapMike is an experienced Product Manager who focuses on all the “non-development” areas of My AskAI, from finance and customer success to product design, copywriting, testing and more.
Table of Contents
- Quick answer: the best AI chatbots for ecommerce pre-purchase questions
- What does an AI chatbot need to answer pre-purchase questions?
- 1. Can it answer sizing and fit questions from your live catalog?
- 2. Can it check stock and give a delivery estimate before purchase?
- 3. Can an AI chatbot recommend products and check compatibility?
- 4. Can it apply a discount code without inventing one?
- 5. Can it add to cart and check out inside the chat?
- 6. Does the same agent handle the order after checkout?
- Shopping assistant or support agent?
- What is agentic commerce, and where is it real?
- How I scored these tools for pre-purchase questions
- The 7 AI chatbots for ecommerce pre-sales: at a glance
- Which of these work on Shopify, and which work outside it?
- Where does pre-sales chat automation fail?
- Failure mode 1: Inventing a promo
- Failure mode 2: Confident answers off a stale catalog
- Failure mode 3: Recommending what you can't ship
- Failure mode 4: Losing the thread at the moment of purchase
- Can My AskAI answer pre-purchase questions?
- How My AskAI handles pre-purchase questions end-to-end
- Capabilities shipped (out of 6)
- Who's using My AskAI for pre-sales?
- How does My AskAI price for pre-sales volume?
- Can Alhena AI answer pre-purchase questions?
- How Alhena handles pre-purchase questions end-to-end
- Capabilities shipped (out of 6)
- Who's using Alhena for pre-sales?
- How does Alhena price for pre-sales volume?
- Can Gorgias AI Agent answer pre-purchase questions?
- How Gorgias handles pre-purchase questions end-to-end
- Capabilities shipped (out of 6)
- Who's using Gorgias for pre-sales?
- How does Gorgias price for pre-sales volume?
- Can Yuma AI answer pre-purchase questions?
- How Yuma handles pre-purchase questions end-to-end
- Capabilities shipped (out of 6)
- Who's using Yuma for pre-sales?
- How does Yuma price for pre-sales volume?
- Can Tidio answer pre-purchase questions?
- How Tidio handles pre-purchase questions end-to-end
- Capabilities shipped (out of 6)
- Who's using Tidio for pre-sales?
- How does Tidio price for pre-sales volume?
- Can Intercom Fin answer pre-purchase questions?
- How Intercom Fin handles pre-purchase questions end-to-end
- Capabilities shipped (out of 6)
- Who's using Fin for pre-sales?
- How does Fin price for pre-sales volume?
- Can Rep AI answer pre-purchase questions?
- How Rep AI handles pre-purchase questions end-to-end
- Capabilities shipped (out of 6)
- Who's using Rep AI for pre-sales?
- How does Rep AI price for pre-sales volume?
- What does answering pre-purchase questions save you? (worked example)
- So which AI chatbot for ecommerce is best for pre-sales?
- FAQs
Created time
Aug 5, 2026 09:54 AM
Title length (<60)
Author
Last optimised
Ecomm?
Tags
Image
Publish date
Jul 30, 2026
Video
Slug
best-ai-chatbot-for-ecommerce-pre-sales-2026
Featured
Type
Article
Ready to Publish
Ready to Publish
We scored 7 AI chatbots for ecommerce on nine pre-purchase criteria. Rep AI wins it on 62 of 90, one point clear of Intercom Fin, and we come third at 59: fifth of seven on capability, and $0.10 a ticket, flat, against the per-resolution pricing of the native helpdesk AIs (on top of the helpdesk seats you keep paying either way). On our own case-study handle times and a $0.40/min loaded rate, about 1,500 pre-purchase conversations a month is roughly $2,400 of agent time by hand.
Cart open, one tab, one thumb: "will this fit a 42 wide, and can I have it by Friday?" Nobody replies for six hours, and by then the shopper has bought it somewhere else or stopped wanting it.
Most stores staff for the queue that arrives after the order, because that queue shouts. The pre-purchase queue just leaves.
I'm one of the founders of My AskAI, so I can't be impartial here, and we're reviewed first on this list. Every tool got scored against the same nine criteria, and we don't come out on top.
One customer we lean on for this job is Edel Optics, who sell designer and own-label glasses into 53 countries, where fit and spec questions are most of the inbox.
Quick answer: the best AI chatbots for ecommerce pre-purchase questions
TL;DR: Rep AI leads for pure in-widget selling, Intercom Fin is the pick if you already run Intercom, and the most predictable line in the set is a flat $0.10/ticket agent. Each of these wins a different setup.
- Rep AI (the conversion-native pick) - on Shopify the whole in-chat path arrives pre-built, from catalog search to add-to-cart, checkout and live Shopify discount codes, with revenue attribution on top; WooCommerce gets a link-based version and Salesforce Commerce Cloud no in-chat apply at all, and you pay per website session, so the bill tracks footfall.
- Intercom Fin (best if you're already on Intercom) - five pre-built Shopify connector templates cover catalog, cart, policies and product detail; a shopping conversation bills like a support ticket: $0.99 per outcome, or $9.99 if Fin logs it as a lead qualification.
- My AskAI (one agent, both halves of the journey) - flat per-ticket pricing on any ecommerce platform, plus a Shopify sales-agent Task that recommends products; the cart and checkout steps are yours to build.
- Alhena AI (best if you're not on Shopify) - the broadest platform coverage here and real agentic checkout; the catalog refreshes once a day.
- Gorgias AI Agent (deepest in-chat selling on Shopify) - the strongest discount engine of the set; Shopping Assistant needs both Shopify and Gorgias, and you get billed twice for the same conversation.
- Tidio (the cheapest way in) - pre-built add-to-cart cards on Shopify and WooCommerce, and stock-aware recommendations; recommendations need a paid Lyro quota and the post-purchase half is a template you still have to wire.
- Yuma AI (post-purchase depth, pre-sales bolt-on) - the strongest returns and order automation in the set; its Sales AI is a separate product with no public docs and no published price.
What does an AI chatbot need to answer pre-purchase questions?
TL;DR: Answering a pre-purchase question takes six distinct capabilities, from live catalog awareness to an in-chat conversion action. Most ecommerce chatbots cover three of them and answer the rest out of your help center.
Answering a pre-purchase question takes six separate wires, and the answer is only as good as the weakest one. Vendor pages compress all six into one line.
1. Can it answer sizing and fit questions from your live catalog?
Variants, sizes, specs, materials, price, read from the store rather than the help center. A help-center-only agent can answer your returns policy all day and still say nothing useful about a 42 wide. We see that question land hardest at Edel Optics and Swytch, where the whole pre-purchase inbox is fit and compatibility.
That takes a catalog connector into Shopify, WooCommerce or BigCommerce, and a sync cadence fast enough to still be true when the shopper asks. Vendors go quiet on the second half: Alhena's docs say it "fetches all products visible to the end user and updates this data daily", so a flash-sale price is a day old in the chat.
2. Can it check stock and give a delivery estimate before purchase?
Two halves: is it in stock right now, and when would it land if I ordered it now? Stock needs a live inventory read; the ETA needs shipping rules or a carrier lookup applied to a basket that doesn't exist yet.
Almost every vendor here does the first half. The second half is where all seven come up short.
3. Can an AI chatbot recommend products and check compatibility?
Narrowing 4,000 SKUs down to the two a shopper should look at, comparing them, and answering "will this work with the thing I already own". That's the job a good shop assistant does.
On our side it's a Shopify Task acting as a sales agent, in the same natural-language Tasks and Tools framework as order lookup.
4. Can it apply a discount code without inventing one?
Current discount codes, live promos, pre-purchase returns and warranty terms, without inventing a code that doesn't exist. Mechanism decides this one: an agent composing from your marketing-page text will eventually make a code up, while one reading a live promo source, or creating a real object in your store admin, can't.
Most "hallucinations" in support are the merchant's own stale docs read back accurately, which is something I wrote about on LinkedIn earlier this year.
5. Can it add to cart and check out inside the chat?
The product card, the add-to-cart button, the checkout link, the code applied for you. Scores on this capability run from 2 to 9, the widest spread of the six.
Some vendors have it switched on out of the box. Others, us included, leave it as something a merchant builds with Tasks and AI Actions.
6. Does the same agent handle the order after checkout?
The same agent picks up the "where is my order" ticket a week later, on the same knowledge and the same bill. Get it wrong and the shopper who just spent nine minutes explaining their setup to a shopping bot explains it again to a support bot.
That takes helpdesk-native deployment plus real order and returns actions, which is why several pure shopping assistants land short.
Shopping assistant or support agent?
Vendors sell two products under one label, and the split decides your shortlist. A shopping assistant (Alhena, Gorgias's Shopping Assistant skillset, Yuma's Sales AI, Rep AI's Sales Agent) reads the catalog first and gets judged on conversion and attributed revenue. An AI support agent (My AskAI, Fin's service role, Tidio's Lyro) is knowledge-first, and its numbers are resolution and CSAT.
The pre-purchase question sits on the seam: does it come in my size, is it in stock, when does it land, will it work with X. Buy one of each and you're paying two subscriptions and maintaining two knowledge bases that drift apart within a quarter.
What is agentic commerce, and where is it real?
Agentic commerce is an AI agent that takes commercial actions on the shopper's behalf inside the conversation. My bar is four of the six capabilities, and below that a tool is only answering questions about your products.
It's real in four places here: Gorgias's in-chat product page with variant selection and add-to-cart, Alhena's agentic checkout, Rep AI's add-to-cart plus in-chat discount apply, and Fin's Get Cart and Update Cart connectors. Everywhere else the label sits on top of a recommender.
I scored all seven, and nobody clears all six. Not one of the seven documents a pre-purchase delivery ETA, so all of them can tell a customer where their parcel is after the order exists, and none can answer "if I order this now, when does it land?" before it does.
How I scored these tools for pre-purchase questions
TL;DR: Every vendor is scored out of 10 against the six capabilities above plus three cross-cutting criteria. Anything covering fewer than four capabilities was left out.
Nine criteria, each out of 10, with Overall as the column sum. Six come straight from the spec above; the other three decide whether a good agent ever gets live and stays affordable: setup work, cost at real volume, and whether it runs on the platform you sell on.
The criteria, in priority order:
- Live catalog awareness
- Real-time stock and delivery ETA
- Guided recommendation and compatibility
- Promo and policy accuracy
- In-chat conversion action
- Post-purchase continuity
- Setup ease
- Cost at typical pre-sales volume
- Ecommerce-platform coverage
Third-party review scores carry no weight here: a star rating measures how a vendor's support team behaves, and I wanted a score for whether its agent can read a variant.
Published resolution and automation rates are out too: any vendor can hit 90% resolution if the escalation button is hard enough to find, which is a point I've made before. Every score here traces to a documented capability instead.
These seven are the names I hear on ecommerce calls. DigitalGenius is post-purchase across all 60-plus of its ecommerce use cases; Crisp, Chatbase and Aissist are general website widgets with Shopify actions bolted on, none doing catalog-aware guided selling.
The 7 AI chatbots for ecommerce pre-sales: at a glance
TL;DR: Rep AI takes the top spot on the pre-built conversion path, Intercom Fin is a point behind for teams already on Intercom, and My AskAI is third, earned on setup, cost and platform reach.
This is the pre-purchase half of the job, scored on pre-purchase criteria.
(scores out of 10) | My AskAI | Alhena AI | Gorgias AI Agent | Yuma AI | Tidio (Lyro) | Intercom Fin | Rep AI |
Live catalog awareness | 7 | 7 | 7 | 3 | 7 | 8 | 8 |
Real-time stock + delivery ETA | 3 | 4 | 4 | 3 | 5 | 5 | 4 |
Guided recommendation + compatibility | 7 | 7 | 8 | 4 | 7 | 8 | 9 |
Promo and policy accuracy | 5 | 3 | 9 | 3 | 3 | 8 | 8 |
In-chat conversion action | 4 | 7 | 8 | 2 | 6 | 8 | 9 |
Post-purchase continuity | 8 | 8 | 8 | 9 | 3 | 8 | 7 |
Setup ease | 8 | 7 | 5 | 2 | 6 | 7 | 7 |
Cost at typical pre-sales volume | 9 | 5 | 2 | 3 | 8 | 3 | 6 |
Ecommerce platform coverage | 8 | 9 | 2 | 3 | 5 | 6 | 4 |
Overall (out of 90) | 59 | 57 | 53 | 32 | 50 | 61 | 62 |
Same nine criteria, in plain words:
(in plain words) | My AskAI | Alhena AI | Gorgias AI Agent | Yuma AI | Tidio (Lyro) | Intercom Fin | Rep AI |
Live catalog awareness | Shopify sync, 24-hour cadence | All products, refreshed daily | Product-level, not variant-level | Merchant-maintained corpus | Shopify API, real-time list | Synced straight from Shopify | Live catalog plus metafields |
Real-time stock + delivery ETA | Stock unconfirmed, ETA post-purchase | Inventory tool, daily refresh | Stock filter in chat only | Tracker is post-purchase | Stock-aware, no pre-purchase ETA | Merchant-set stock filter | Stock filtered, not answered |
Guided recommendation + compatibility | Shopify sales-agent Task | Quiz, outfit builder, narrowing | In-chat grids and product pages | Behavior nudges, no narrowing | Eight conversation types | Narrows and compares thousands | Product Finder, no rigid quiz |
Promo and policy accuracy | Guidance and Custom Answers | Guideline-level control only | Real codes created in Shopify | Generates codes, no guard | Coupons live in Flows | Shop policies connector | Codes pulled live from Shopify |
In-chat conversion action | Buildable with Tasks and Tools | Pre-built on Shopify only | Card, add-to-cart, product page | Nudges, no add-to-cart | Add-to-cart, Shopify and Woo | Five pre-built Shopify templates | Cart, checkout, in-chat codes |
Post-purchase continuity | Same agent, same ticket bill | Same agent, same credit pool | One agent, two skillsets | The core product, end to end | Gated template, custom Actions | Procedures for returns, refunds | Returns initiated, not executed |
Setup ease | Self-serve, 30-day trial | Self-serve, free tier | Gorgias helpdesk required | Sales call before anything | Cheap entry, three gates | One-click Shopify, theme embed | App Store install, SDK extras |
Cost at typical pre-sales volume | $0.10 a ticket, flat | Credits, then $1.20 overage | Billed twice, $1.50 overage | Unpublished, sales-gated | Low entry, three meters | $0.99 every outcome | Metered on site visitors |
Ecommerce platform coverage | Any platform via the helpdesk | Broadest: Magento, BigCommerce, SFCC | Shopify and Gorgias only | Sales AI is Shopify-only | Shopify, Woo, OpenAPI elsewhere | Shopify native, connector elsewhere | Shopify-first, others degraded |
Three vendors tie at five of six capabilities: Gorgias, Intercom Fin and Rep AI. They separate on the cross-cutting rows, where Rep AI's session meter, Fin's per-outcome rate and Gorgias's double bill get priced in.
On the six capability rows alone, out of 60, it reads Rep AI 45, Intercom Fin 45, Gorgias 44, Alhena 36, My AskAI 34, Tidio 31, Yuma 24. So we sit fifth of seven on raw capability.
Third overall comes down to setup, cost and platform coverage. If your only question is "which of these does the most inside the widget", the answer isn't us.
Which of these work on Shopify, and which work outside it?
Shopify-only is more common here than the marketing suggests. Gorgias AI Agent is the sharpest case: the helpdesk supports WooCommerce, BigCommerce and Magento, but the eligibility box on all four Shopping Assistant docs reads "Can be used with Shopify stores connected to Gorgias".
Yuma's Sales AI is Shopify-only too, and Tidio's add-to-cart covers Shopify and WooCommerce, with a platform-agnostic OpenAPI route that's excluded from the free trial.
Intercom Fin is native on Shopify and, in its own ecommerce docs, "can also be deployed on other platforms (WooCommerce, Magento, BigCommerce, Salesforce Commerce Cloud, and custom-built stores) via a catalog connector", which isn't native sync. Rep AI is Shopify-first with WooCommerce and Salesforce Commerce Cloud degraded. Alhena has the widest documented reach, with catalog support on Magento 2, BigCommerce, Salesforce Commerce Cloud, Solidus and spreadsheet import.
Scoreboard order, and what each one actually does on the two sides of that line:
Vendor | On Shopify | Outside Shopify |
Rep AI | Full experience, in-chat code apply | WooCommerce link-based, SFCC degraded |
Intercom Fin | One-click connect, native sync | Catalog connector, not native sync |
My AskAI | The one pre-built connector | Any platform, through your helpdesk |
Alhena AI | Add-to-cart native, no code | Magento 2, BigCommerce, SFCC, Solidus |
Gorgias AI Agent | Needs Shopify plus the Gorgias helpdesk | Shopping Assistant unavailable |
Tidio | Add-to-cart on product cards | WooCommerce, then OpenAPI elsewhere |
Yuma AI | Sales AI runs here only | Sales AI unavailable |
We run through your helpdesk, so our agent is platform-agnostic, with Shopify as the one pre-built connector. If you're on Shopify, we've written a separate walkthrough of AI customer service on Shopify.
Where does pre-sales chat automation fail?
TL;DR: Four failure modes kill pre-sales chat: inventing a discount code, confident answers off a stale catalog, recommending something you can't ship, and dropping the thread the moment the shopper becomes a customer.
Merchants have been skeptical of this category for years, mostly for good reasons. Here's a store owner and store builder on the Shopify Community back in December 2023:
"Speaking as a store owner, a person who builds stores for others, and someone who shops online a lot – I really don't like AI chat bots. The overwhelming majority of the time that I encounter them, they don't actually solve the issue I have – or I've already read the help docs that they provide, and I just want to speak to a real person."
That's the objection the spec exists to answer. It still goes wrong four ways in 2026.
Failure mode 1: Inventing a promo
The agent offers a student discount on a store that has never run one, the shopper screenshots it, and you honor it to save the sale. Root cause is component 4 left unwired: the agent is composing from marketing-page text instead of reading a live promo source.
Gorgias can't do this. Its discount docs describe Shopping Assistant creating "unique, single-use discount codes", and specifically that "it creates a new unique discount code in your Shopify admin", each one "valid for 3 days and limited to a single use" (three days is short enough that a screenshotted code is dead by the time it circulates).
Every code is a real object inside your own store, so there's nothing to invent. Rep AI takes the import route instead, pulling existing Shopify codes into chat.
Alhena's control is guideline-level only, and its own docs offer the example rule "Never offer discounts unless the customer has a valid promo code." Tidio's coupons live in Flows, a separately metered product.
My disqualifier: ask "do you have a student discount?" on a store that doesn't, and watch whether it invents one.
Failure mode 2: Confident answers off a stale catalog
Nightly or manual sync means the agent is answering accurately about yesterday's price and yesterday's stock. Alhena refreshes once a day, so a reprice is wrong in the widget until the next sync.
Tidio reads Shopify's product schema. Its docs say "It doesn't scrape your individual product's webpage", so third-party variant, currency and stock apps are unreadable, and on some stores it fails silently: "For a (small) number Shopify stores, the feature will not display due to an incompatibility with your Shopify schema." Yuma puts the burden on you outright, instructing merchants to "Keep variant data and inventory in sync with the page".
My disqualifier: get the catalog sync cadence in writing.
Failure mode 3: Recommending what you can't ship
Guided selling with no inventory check is component 3 running without component 2. It manufactures intent and then kills it. A merchant on the Shopify Community in November 2024 priced it for us:
"I've run into some problems with my store. After running some ads and getting no conversions did some research. Turns out that both ChatGPT and the Shopify AI bot said that the product is "out of stock", even though I've got plenty left in my inventory on Shopify."
Paid traffic, landing on a bot that says the thing isn't available. Tidio guards this best of the set: "Lyro is also aware of your stock, so it won't recommend items that are currently unavailable (or it will clearly mention that the item is out of stock)."
Gorgias guards it in chat only, and its docs say that on email and SMS it may still discuss unavailable products. Fin makes it a merchant setting called Configure stock filter, warning that for made-to-order or availability-restricted products it "may match on the wrong product page and give customers inaccurate availability information".
My disqualifier: ask whether the recommender is stock-aware, and get the answer out of the docs.
Failure mode 4: Losing the thread at the moment of purchase
The pre-sales bot and the support agent are different products, so the buyer who just had a nine-minute sizing conversation re-explains everything on their first WISMO ticket, and you pay two vendors for the privilege. Yuma is the structural version: Sales AI and Support AI are separate products with separate, unpublished pricing.
Tidio's post-purchase half is a template you still wire, gated to Plus, Premium and selected Lyro subscribers. Gorgias's Ecom Lab benchmarks, updated May 2026 across 1,000-plus ecommerce brands, put the median wait between AI handoff and human response at 10 hours, with 33% of handed-off tickets abandoned.
My disqualifier: ask what happens to the conversation after checkout. Same agent, same knowledge, same bill, or a second product and a second invoice?
Can My AskAI answer pre-purchase questions?
TL;DR: We cover 3 of 6 outright with 3 partial, on any ecommerce platform, inside the helpdesk you already run. Strength: one agent doing pre- and post-purchase at a flat $0.10 a ticket. Weakness: the in-chat conversion action is a build with Tasks and Tools.
On our side, a Shopify Task acts as a sales agent, offering product recommendations in the chat. It sits in the same natural-language framework as order lookup and refunds, on the same bill, so recommendation isn't a second subscription with a second knowledge base behind it.
Our position is pre-sales support first, selling second: most of what a shopper asks before buying is a support question in disguise. The pure shopping assistants have built further down the conversion path, and they score higher for it.
How My AskAI handles pre-purchase questions end-to-end
Knowledge comes in from your help center, website, files, past tickets and connectors. Our Shopify integration connector syncs products, pages and customer data on a 24-hour cadence, and it works with Shopify Markets for multi-language, multi-currency and multi-store setups. Live order data on any other platform comes through the User Data API.
Recommendations run as a Task. Tasks are multi-step workflows written in plain language, and the agent "asks the right questions at the right time", so a sizing conversation branches the way a person would.
Compatibility comes off the same layer: synced site pages, Custom Answers for exact wording, Guidance rules for scenarios you want handled a specific way. Out of the box that Task recommends and stops there, which is why we score a 4 on the conversion row.
Add-to-cart, a checkout link and applying a discount code are buildable with Tasks plus AI Actions, which support "any API request method, including POST, GET, PATCH, PUT and more", but they don't arrive configured. If you want the cart populated from chat on day one without a build, Rep AI, Gorgias and Fin get you there faster.
The Order Status Check Task and Tool are "automatically created for you on account set up", and returns, refunds and cancellations run through the same framework, autonomously or as propose-then-approve, set per action. When you want to know why the agent said what it said, ask Echo which knowledge source it used.
Capabilities shipped (out of 6)
Capability | Shipped? |
Live catalog awareness | ✅ Shopify connector syncs products, pages and customer data; 24-hour cadence on most sources |
Real-time stock and delivery ETA | ⚠️ Order data covers tracking, delivery status and estimated delivery dates once the order exists; no documented pre-purchase ETA, the same partial as every other tool here |
Guided recommendation and compatibility | ✅ Shopify sales-agent Task inside Tasks and Tools, on the same knowledge as the rest of the agent |
Promo and policy accuracy | ⚠️ Custom Answers return exact wording and Guidance constrains scenarios, but there's no documented live promo feed read at answer time |
In-chat conversion action | ⚠️ Recommends out of the box; add-to-cart, checkout links and code application are buildable with Tasks and AI Actions, not pre-built |
Post-purchase continuity | ✅ Same agent, same helpdesk, same per-ticket bill; order status auto-created, returns and refunds via Tasks |
Who's using My AskAI for pre-sales?
Edel Optics is the closest match to this job in our customer base: a European eyewear retailer on Zendesk selling into 53 countries, where fit and spec questions are the bulk of the inbox. They run 75-79% AI resolution across 4,067 tickets at 92% CSAT, up from 20-30% before live customer data went in.
YouGarden, a UK online garden center on Freshdesk, is the pre-purchase advice case: the agent answers plant-care questions before the sale off 3,000-plus synced pages spanning their own site, the RHS site and their care-guide site, at 66% resolution and 78% CSAT across 11,785 tickets. Their head of customer service put it like this:
"What impressed us most was how accurately the AI reflects our tone, policies, and product knowledge. The responses feel genuinely helpful rather than generic, which is critical in a sector like horticulture where advice really matters."
Swytch sells e-bike conversion kits on Zendesk, where almost every pre-purchase question is a version of "will this fit my bike?".
We count a conversation resolved when the AI handled it without escalating to a human, with escalation always one click away. So the number describes what didn't need a person, and CSAT is the separate check on whether they were happy.
Our field benchmark across 195 rated deployments and 38 vendors puts the median around 70%, our own rolling rate at 72%. Those figures are directional, because every vendor's numerator is defined differently.
How does My AskAI price for pre-sales volume?
We charge flat, per ticket. Scale is $499 a month covering 2,000 tickets, then $0.10 each after that; Pro is $199 covering 1,000, at roughly $0.12 each up to a 5,000 ceiling.
At 1,500 pre-purchase conversations a month you're inside the Scale allowance at about $499, on top of the helpdesk seats you pay anyway. Two add-ons apply: Tasks are $0.02 a step and AI Actions $0.02 per reply where a tool call fires, so a recommendation Task carries a few cents on top of the ticket.
The full ladder is on the pricing page, and the trial runs 30 days with every feature unlocked, unlimited tickets and no card.
Choose My AskAI for pre-purchase questions if:
- You want one agent covering the pre-purchase question and the WISMO ticket that follows it, on the same knowledge and the same invoice
- You sell on something other than Shopify, or on more than one platform, and want the agent to work the same way everywhere
- You want a bill that stays flat as the agent gets better and the conversions climb
Our Shopify integration and Tasks and Tools pages cover the recommendation pattern, and the pricing page has the per-ticket math.
Can Alhena AI answer pre-purchase questions?
TL;DR: 4 of 6 covered with 2 partial. Strength: the only documented agentic checkout in the set, and the broadest platform reach. Weakness: the catalog refreshes once a day, and add-to-cart is only pre-built on Shopify. From $199 a month on annual terms, $239 monthly.
Alhena is the purest shopping assistant here. If you sell somewhere other than Shopify, it's where I'd look first. It's catalog-first, conversion-measured, and it reaches platforms nobody else in this set documents.
How Alhena handles pre-purchase questions end-to-end
The agent answers with cards carrying image, title, price and rating, then narrows conversationally: ask for snowboarding gear for a beginner and it returns a priced grid and offers to explain the differences. Four guided surfaces sit on top, documented as embeddable agents: a product quiz, an outfit builder, a room designer and virtual try-on (the outfit builder is sales-gated, with the docs telling you to "Contact the Alhena team to enable the Outfit Builder for your account").
Conversion is where the platform caveat bites. Alhena's add-to-cart docs put Shopify first: "No custom code required. The Alhena widget automatically detects Shopify stores and handles add-to-cart natively."
Everywhere else, the button doesn't appear on product cards until a developer registers a
products:added_to_cart event handler. Agentic checkout is real though, and the only one I found in this set: Tatcha's case study describes the agent filling the cart and redirecting to a pre-populated checkout.Promos are the weak wire, and post-purchase tracking is a third-party hop through Ship24, Narvar or Loop.
Capabilities shipped (out of 6)
Capability | Shipped? |
Live catalog awareness | ✅ with a caveat: "fetches all products visible to the end user and updates this data daily" |
Real-time stock and delivery ETA | ⚠️ inventory checks are a listed API-tool use case, but the catalog is a day old and no delivery ETA is documented |
Guided recommendation and compatibility | ✅ for narrowing (quiz, outfit builder, conversational search); ⚠️ no documented "will this work with X" check |
Promo and policy accuracy | ⚠️ guideline-level only, e.g. "Never offer discounts unless the customer has a valid promo code." No code generation or validity check |
In-chat conversion action | ✅ pre-built on Shopify, ⚠️ developer work elsewhere; agentic checkout documented |
Post-purchase continuity | ✅ same agent and same credit pool for tracking, returns and exchanges |
Who's using Alhena for pre-sales?
Tatcha, the Unilever Prestige beauty brand, is the flagship: 3x site-average conversion, a 38% AOV increase and 11.4% of total site revenue attributed, with product cards, agentic checkout and adaptive quizzing named as the features doing it. Their director of digital product and design put the brief like this:
"Our goal isn't just automated support; we need an AI that can sell, educate, and retain—providing expert product recommendations, building skincare routines, and delivering post-purchase guidance."
Victoria Beckham is the other published pre-sales name, where the agent handles product-specific questions on sizing and materials.
How does Alhena price for pre-sales volume?
Alhena bills in conversation credits, and I'd read the definition before the tiers: the pricing page counts one credit "when Alhena's AI resolves or assists a customer interaction", with spam or immediate human handoffs costing nothing. Free covers 25 conversations a month; Essentials is $199 annual or $239 monthly for 200; Growth is $499 or $599 for 550; Scale is $999 or $1,199 for 1,200.
Overage is $1.20 a credit on every tier, with no seat fees. At 1,500 pre-purchase conversations you're on Scale plus 300 credits of overage, so roughly $1,359 a month on annual terms.
Choose Alhena for pre-purchase questions if:
- You sell on Magento 2, BigCommerce or Salesforce Commerce Cloud and want catalog support that's documented
- You want agentic checkout, with the cart filled and checkout pre-populated from the conversation
- Guided surfaces (quiz, outfit builder, try-on) are a real part of how you sell
For more on Alhena, read our complete guide, the Alhena alternatives roundup or the head-to-head against My AskAI.
Can Gorgias AI Agent answer pre-purchase questions?
TL;DR: 5 of 6 covered with 1 partial, joint-highest capability coverage in the set. Strength: a discount engine that can't hallucinate a code, plus a full product page inside the chat. Weakness: Shopping Assistant needs Shopify and Gorgias, and you're billed twice for the same conversation. From $40 a month bundled, $1.50 per automated interaction past your allotment.
Gorgias has built the most complete in-chat selling experience of anything I looked at, and hemmed it in with the tightest prerequisites. If you're already a Gorgias store on Shopify, this is the deepest option on the list.
How Gorgias handles pre-purchase questions end-to-end
One AI Agent carries two skillsets, Support and Shopping. The shopping half reads your Shopify catalog ("product names, descriptions, and main product image") plus your synced store site, and recommends at the product level rather than the variant level, so "do you have this in a 9?" isn't a recommendation-engine question there, though variant selection exists once the shopper is on a card.
That card is the strongest conversion surface here: a visual Shopify product card with direct add-to-cart and no merchant code, opening into a full in-chat product page with color swatches, a size selector and a quantity stepper. Email and SMS fall back to a text link.
Selling behavior is configurable across four discount strategies and three selling styles, more control than anyone else here exposes and more than I'd switch on in week one. Discounts are the standout: the agent creates a real single-use code inside your Shopify admin, valid for three days, and it detects your four Shopify automatic-discount types so it doesn't stack something it shouldn't.
Stock awareness is region-based and applies "in chat conversations" only. On the bill, Gorgias's own billing docs state that "You are charged both a ticket fee + automation fee if AI Agent responds to a ticket and does not hand over the conversation to a human agent".
Capabilities shipped (out of 6)
Capability | Shipped? |
Live catalog awareness | ✅ Shopify catalog plus synced site and custom details; recommends at product level, not variant level |
Real-time stock and delivery ETA | ⚠️ won't recommend region-unavailable products in chat, but may still discuss them on email and SMS; no pre-purchase ETA |
Guided recommendation and compatibility | ✅ in-chat grids and product pages, prioritized by browsing activity and buying-intent stage |
Promo and policy accuracy | ✅ strongest in the set: real single-use codes created in your Shopify admin, valid three days, aware of automatic discounts |
In-chat conversion action | ✅ pre-built visual product card with add-to-cart and a full in-chat product page |
Post-purchase continuity | ✅ one agent, two skillsets: cancellations, returns, refunds and subscription edits |
Who's using Gorgias for pre-sales?
TUSHY is the compatibility case study of the set, using Shopping Assistant for does-this-bidet-fit-my-toilet guidance, installation reassurance and recommendations, and reporting a 15% lift in chat conversion rate, conversion at twice the rate of human agents, and 13x ROI (their own published figures).
Arc'teryx reports 23x ROI on AI Agent and a 75% conversion lift from Shopping Assistant, and bareMinerals is also published.
How does Gorgias price for pre-sales volume?
The old standalone per-resolution add-on is gone, replaced by bundled helpdesk-plus-AI tiers on the pricing page: Starter $40 (50 tickets, 30 automated interactions), Basic $90 (300/30), Pro $550 (2,000 tickets but only 190 automated interactions) and Advanced $1,430 (5,000/530).
Comparing AI charge to AI charge, the in-bundle rate works out at $0.90 to $1.00 an interaction, and anything past the allotment is $1.50, a 67% premium. At 1,500 automated pre-purchase conversations on Pro that's roughly $550 plus about $1,965 of overage, with the ticket fee applying to the same conversations.
Starter has no Actions and no brand-voice controls, so the $40 plan can't apply a discount or cancel an order.
For more on Gorgias, read our complete guide to its AI Agent, the Gorgias Automate alternatives roundup, or the comparison with My AskAI.
Can Yuma AI answer pre-purchase questions?
TL;DR: 1 of 6 covered with 5 partial, the thinnest pre-sales coverage here and the deepest post-purchase one. Strength: returns, refunds and subscription automation done end to end. Weakness: Sales AI is a separate bolt-on with no add-to-cart, no public documentation and no published price. Pricing is sales-gated.
Yuma is a strong post-purchase product with a pre-sales widget attached, and the two are sold separately. Yuma documents almost none of the pre-sales half, so on most capabilities there is no evidence either way.
Yuma publishes no help center at all, so a buyer can't self-assess before a call.
How Yuma handles pre-purchase questions end-to-end
Sales AI installs as a JavaScript snippet on a Shopify store and answers on the product page. Yuma describes it as reading the live store: "Yuma's AI shopping agent reads your live product catalog, reviews, and CRM in real time, so every recommendation is specific to your actual store."
Its own buyer-question guide describes retrieval from a vetted corpus of product specs, size charts and policy docs, and tells merchants to keep variant data and inventory in sync with the page, which reads to me more like a maintained corpus than a live read.
Recommendations are behavioral: Smart Recommendations and Next-Best Buy, driven by browsing, cart and purchase history. Sizing, compatibility and comparison are named as target question categories, but no guided-narrowing or SKU-versus-SKU flow is documented anywhere.
Promos run through a Custom Coupon Generator producing a personalized code validated by something called Safety Guard, with no documented check against what's live on the store. That's failure mode 1 waiting to happen, and I found nothing in Yuma's public material that rules it out.
Conversion actions are the gap: Affinity Nudges push shoppers "toward product details, recommendations, or limited-time deals" and Cart Saver recovers abandoned carts, but no add-to-cart, checkout link or code application is documented. Support AI handles order tracking, returns, refunds and subscription management end to end inside your existing helpdesk.
Capabilities shipped (out of 6)
Capability | Shipped? |
Live catalog awareness | ⚠️ marketing describes a real-time read; the documentation describes a merchant-maintained corpus you keep in sync |
Real-time stock and delivery ETA | ⚠️ not documented for Sales AI; real carrier data exists only in the post-purchase Package Tracker |
Guided recommendation and compatibility | ⚠️ behavior-driven Smart Recommendations and Next-Best Buy; no documented narrowing or comparison flow |
Promo and policy accuracy | ⚠️ Custom Coupon Generator plus Safety Guard, with no documented guard against surfacing a code that isn't live |
In-chat conversion action | ⚠️ nudges and cart recovery only; no add-to-cart, checkout link or code application documented |
Post-purchase continuity | ✅ the core product: tracking, returns, refunds and subscriptions, inside your existing helpdesk |
Who's using Yuma for pre-sales?
Yuma publishes no dedicated pre-sales case study. Its case-study index lists Tediber, Glossier, Petlibro, Clove, Cabaïa, FINN and others, and every one is support or post-purchase.
The single pre-sales name it publishes anywhere is Parachute Home, which appears only in the Sales AI launch post reporting an 18% lift in revenue per visitor and 4% higher AOV from an A/B test (one test, with no dedicated study behind it).
How does Yuma price for pre-sales volume?
You can't price it. The pricing page publishes no figures at all, offering "Talk to us for pricing details." under a performance-based-billing headline, and the mechanism is described on the Support AI page as "Performance-based pricing means you only pay for tickets the AI fully resolves, with a 100% ROI guarantee on every contract."
Nothing is published for Sales AI specifically. Setup is sales-gated by design, with the FAQ stating that "A dedicated account manager from Yuma configures your AI automations", and there's a 30-day trial on your real ticket data, which is the way to get a number.
Choose Yuma for pre-purchase questions if:
- Your real problem is post-purchase volume (returns, refunds, subscriptions) and pre-sales is a bonus
- You run Gorgias, Zendesk, Kustomer, Gladly, Front or Re:amaze and want deep Shopify actions inside it
- You're happy to buy through a sales process with a dedicated account manager doing the configuration
For more on Yuma, read our complete guide or the Yuma AI alternatives roundup.
Can Tidio answer pre-purchase questions?
TL;DR: 3 of 6 covered with 3 partial. Strength: pre-built add-to-cart cards and stock-aware recommendations at the lowest entry price in the set. Weakness: recommendations need a paid Lyro quota, the post-purchase half is a gated template, and it reads Shopify's schema, so anything outside it is invisible. Lyro starts at $32.50 a month.
Tidio's Lyro is the default widget on a lot of smaller Shopify stores, and on pre-purchase it does more than I expected for the price. The limits are in what it can read and what it's allowed to do.
How Tidio handles pre-purchase questions end-to-end
Lyro imports your Shopify products through the API, variants included. Where it trips is the Shopify product schema: "It doesn't scrape your individual product's webpage", so custom blocks outside the schema are invisible and third-party variant, currency and stock apps can't be read at all.
The conversation is well specified. Eight documented types cover product comparison, variant recommendation, alternative recommendation, complementary recommendation and price-range recommendation, and Lyro "will be proactive, asking questions that help find what the customer is looking for." Complementary recommendations only work if the relationship is written in the product description, so I'd check your descriptions before counting on it.
The conversion action is pre-built: each product card carries an add-to-cart button on Shopify and WooCommerce stores, and clicking it lists the available variants (you can flip that to View product in the Sales strategy settings). There's no checkout link or code application, and discount codes live in Flows, a separate metered product.
Everywhere else you're on the OpenAPI route, which works with any ecommerce backend, custom CMS included, but isn't available during the free trial.
Post-purchase is thinner. Order-status templates are gated to Plus and Premium subscribers plus selected Lyro subscribers, and returns and refunds need custom Lyro Actions that are "meant for advanced users or developers in general", and the docs warn that "Lyro does not validate or roll back your integrations. Test in a safe environment; unintended changes to external systems are irreversible by Tidio."
Capabilities shipped (out of 6)
Capability | Shipped? |
Live catalog awareness | ✅ Shopify API import with variants, updated as products change; reads the schema, not your page |
Real-time stock and delivery ETA | ⚠️ stock is covered: "Lyro is also aware of your stock, so it won't recommend items that are currently unavailable"; no pre-purchase delivery ETA |
Guided recommendation and compatibility | ✅ eight documented conversation types including comparison and variant recommendation; compatibility only if it's in the description |
Promo and policy accuracy | ⚠️ policies come from the knowledge base; discount codes are a Flows action, a separately metered product |
In-chat conversion action | ✅ pre-built add-to-cart on product cards, Shopify and WooCommerce only; no checkout link or code application |
Post-purchase continuity | ⚠️ order-status template is gated, returns and refunds need developer-built Actions with no rollback |
Who's using Tidio for pre-sales?
Burker, an Amsterdam D2C watch and jewelry brand on Shopify running around 450,000 monthly sessions, is the most exact pre-purchase description any vendor publishes: Lyro handles sizing, shipping timelines, the difference between a standard model and its Petite counterpart, and which pieces work together as a set, at an 80% resolution rate across 10,000 conversations a month.
Burker has deliberately not switched the product-recommendation layer on yet, wanting their Shopify product data cleaned up first. Suitor, an Australian suit and tux business, resolves 85% of queries and cut response time from three minutes to six seconds:
"Lyro gives customers 24-hour support. Even if it doesn't always fully answer the question, it gives them enough information to go through the website."
How does Tidio price for pre-sales volume?
Three separate meters, and I'd model all three before you buy: billable human conversations, Lyro conversations and Flows visitors. On the pricing page, Free is $0, Starter $24.17 a month, Growth from $49.17, Plus from $300 a month plus usage, and Premium is quoted.
The Lyro add-on starts at $32.50 a month from 50 conversations, and a Lyro conversation is any interaction with at least one AI reply, however long it runs. At 1,500 pre-purchase conversations a month you're into Plus territory, so from $300 plus Lyro usage on top.
Product recommendations key off a paid Lyro quota, so the plan tier alone won't unlock them, and the trial runs seven days.
For more on Tidio, read our complete guide to Lyro, the Tidio Lyro alternatives roundup, or the comparison with My AskAI.
Can Intercom Fin answer pre-purchase questions?
TL;DR: 5 of 6 covered with 1 partial, joint-highest capability coverage. Strength: five pre-built Shopify connector templates covering catalog, cart, policies and product detail, with the same agent handling post-purchase. Weakness: a shopping conversation bills like a support ticket at $0.99 per outcome, or $9.99 if Fin logs it as a lead qualification, and the ecommerce role is Shopify-and-Messenger-shaped. $0.99 per outcome.
Fin for Ecommerce is a distinct product surface from the Fin most people have tried. If you're already anchored on Intercom, this is where I'd start.
How Intercom Fin handles pre-purchase questions end-to-end
Setup is a Shopify one-click connect plus an app embed into your theme (the embed is the bit that rules out a standalone Zendesk deployment). Pricing, availability, variants, stock and product page content then sync straight from your Shopify store and stay current as it changes.
Intercom's explainer describes an agent that "asks the right questions, narrows options from thousands of products, and compares them based on what the shopper actually needs." Fin decides shopping-versus-support itself, with no routing for you to build, and tags ecommerce conversations automatically with Product recommended, Cart created and Checkout link clicked.
Actions come from five pre-built Shopify connector templates: Search Shop Catalog, Get Cart, Update Cart, Search Shop Policies And FAQs, and Get Product Details. Update Cart covers adding and removing line items, buyer information, shipping details, discount codes, gift cards and notes in a single call, which is why component 5 scores as well as it does.
Policies come straight from your Shopify store policies, and stock is a merchant setting called Configure stock filter (I'd test it against your own catalog before you switch it on). Post-purchase Procedures cover order tracking, returns, refunds, exchanges and cancellations, though a new or incomplete one arrives in draft and only runs once you review each step.
Capabilities shipped (out of 6)
Capability | Shipped? |
Live catalog awareness | ✅ native Shopify sync of products, variants, pricing, availability and page content |
Real-time stock and delivery ETA | ⚠️ merchant-set stock filter with an accuracy warning on made-to-order items; no pre-purchase delivery ETA |
Guided recommendation and compatibility | ✅ narrows from thousands of products and compares against stated needs |
Promo and policy accuracy | ✅ Search Shop Policies And FAQs reads your store policies; cart calls carry discount codes and gift cards |
In-chat conversion action | ✅ five pre-built connector templates including Get Cart and Update Cart, live to test instantly |
Post-purchase continuity | ✅ returns, refunds and order changes resolved in the same conversation via Procedures |
Who's using Fin for pre-sales?
WHOOP is the sharpest published pre-purchase deployment in this set, because of where they put it: the Join page, the final step before purchase. I'd steal that placement. Intercom reports a 130% increase in attributed sales, an 84% Fin resolution rate and 3,500-plus resolutions a month, with Fin absorbing thousands of buyer questions during a product launch and resolving 68% without escalation.
Their customer experience lead described the problem it solved:
"If someone chatted with us live, they were twice as likely to convert. But when we couldn't respond fast enough, we lost the moment."
How does Fin price for pre-sales volume?
$0.99 per outcome, with a 50-outcome monthly minimum if you run Fin on a non-Intercom helpdesk and no seat costs for teammates. On the pricing page, running it with the Intercom helpdesk adds $19 per seat per month, with Lite seats free.
You're charged one outcome per conversation however many actions Fin takes, and the billable outcomes are resolution, procedure handoff and disqualification at $0.99, with a qualification at $9.99. At 1,500 pre-purchase conversations that all resolve, comparing AI charge to AI charge, that's roughly $1,485 a month on top of the helpdesk seats you keep paying either way.
The trial is 14 days with unlimited outcomes and no card.
For more on Fin, read our complete guide to the Fin AI Agent, the Intercom Fin alternatives roundup, or the comparison with My AskAI.
Can Rep AI answer pre-purchase questions?
TL;DR: 5 of 6 covered with 1 partial, and the most conversion-native product here. Strength: the whole in-chat path is pre-built, from catalog search to add-to-cart, checkout and live Shopify discount codes, with revenue attribution over the top. Weakness: you pay per website session, three pre-sales controls sit behind $200-a-month add-ons, and SOC 2 is an audit in progress. From $49 a month for Support or $104 for Sales at 10,000 sessions.
Rep AI tops the scoreboard, the only vendor here built as a sales tool first. One G2 reviewer, an ecommerce manager, names the trade-off:
"You need to keep in mind that Rep AI is a sales tool. Our mistake was counting on it just as a customer support bot of sorts."
How Rep AI handles pre-purchase questions end-to-end
Products, variants, collections and Shopify metafields sync into a catalog the agent searches live, with the marketing site claiming it "searches your live Shopify catalog in real time". There are ceilings: 20,000 products, five metafields maximum, and product-level controls.
Guided selling is the strongest of the seven. Product Finder and AI-Powered Collection Search interpret intent, ask clarifying follow-ups and narrow results dynamically without rigid quiz logic. Compatibility is demonstrated on Rep AI's own site, where the flow answers "will this fit an Apple Studio Display?" with a yes and an add-to-cart (the closest thing to a live proof any vendor here offers).
Add-to-cart is a per-product toggle in the console, and applying a discount code without leaving the chat shipped in late July 2026, with codes straight out of Shopify: the docs say the agent "pulls discount details live from Shopify" with no sync delay.
The same page grades the experience by platform (the clearest platform statement any vendor here makes): "Shopify stores get the full experience", "WooCommerce stores get a link-based version", and "Salesforce Commerce Cloud stores don't support in-chat apply."
Stock is filtered out of the results: an out-of-stock item returns a grayed card and a "Notify me" capture routed to Klaviyo, with sync guidance telling you to allow 24 hours. Post-purchase runs nine support skills on the same agent, though returns and exchanges stop at initiated.
Capabilities shipped (out of 6)
Capability | Shipped? |
Live catalog awareness | ✅ products, variants, collections and metafields synced and searched live; caps at 20,000 products and five metafields |
Real-time stock and delivery ETA | ⚠️ stock is filtered rather than answered (grayed card plus a Klaviyo notify-me, sync allow 24 hours); no pre-purchase delivery ETA |
Guided recommendation and compatibility | ✅ Product Finder and collection search, narrowing dynamically without quiz logic; compatibility demonstrated |
Promo and policy accuracy | ✅ real Shopify codes pulled live with no sync delay |
In-chat conversion action | ✅ add-to-cart, checkout and in-chat discount apply, graded by platform |
Post-purchase continuity | ✅ nine support skills on the same agent, but returns and exchanges are initiated, not executed |
Who's using Rep AI for pre-sales?
Twenty-four published case studies, and the pre-sales ones are the substantial ones: Bikes Online attributes 8.4% of revenue to it with AOV up 48%, and Vertical Spice reports converting 21% of shoppers and 9% attributed revenue. OLLY (96% resolved, 99% CSAT) and VIBAe (97% resolution across 150-plus countries) are the headline resolution numbers. The index page only surfaces seven of the 24, so go to the individual study.
Take those resolution figures with a grain of salt. Published vendor numbers are self-selected marketing wins, "answered" is a different numerator from "resolved", and our own field benchmark across 195 rated deployments puts the market median nearer 70%. Rep AI's 97-99% sits well outside that distribution, which usually means the metric is defined differently.
Rep AI's platform page says "SOC 2 compliant", while its trust portal shows an outstanding SOC 2 system-description task and publishes only a gated engagement letter. That's an audit in progress, and a direct question to ask if your reviewer cares.
How does Rep AI price for pre-sales volume?
Metered on monthly website sessions rather than conversations, resolutions or seats, which makes it the odd one out in this set. On the rendered pricing page, AI Sales runs $104 at 10,000 sessions, $209 at 25,000, $368 at 50,000 and $1,355 at 250,000; AI Support is cheaper at each tier ($49, $79, $149, $575) and AI Concierge bundles both, with yearly billing taking 20% off.
A separate Shopify App Store ladder exists with its own numbers and a free-to-install tier at 100 visitors a month, so always check which surface a quoted figure came from. Three pre-sales controls (Flow Studio Builder, Custom Support Policies and Product Finder Customization) are $200 a month each unless you buy Concierge, and Product Finder isn't available on the Support plan at all.
At 1,500 pre-purchase conversations the session meter is the wrong unit to model with, so price it off your traffic. One merchant review on the Shopify App Store:
"shut me off after 8 months and a certain number of views… not worth $550 per month for a basic LLM"
Choose Rep AI for pre-purchase questions if:
- You sell on Shopify and want the full conversion path (search, cart, checkout, live codes) working without a build
- You want revenue attribution on the chat, with CVR and AI-generated sales in the console
- Your traffic is predictable enough that a session-based bill is easy to forecast
For the wider ecommerce picture, read our roundup of the best AI customer service tools for ecommerce.
What does answering pre-purchase questions save you? (worked example)
TL;DR: At around 1,500 pre-purchase conversations a month, the labor bill is roughly $2,400 and the tools range from about $300 to about $2,500. The bigger number is the revenue on the sessions that only convert because someone answered.
A pre-purchase question takes about four minutes of agent time. That comes from our own published case studies: Edel Optics' case-study math uses three minutes a ticket, YouGarden's uses five, so four is the midpoint.
The loaded rate is $0.40 a minute. US Bureau of Labor Statistics wage data puts the mean customer-service-representative wage at $22.40 an hour (about $0.37 a minute before any benefits), and the BLS compensation series puts wages at 69.9% of total compensation, which loads it to about $0.53.
$0.40 sits between bare wage and fully loaded, giving roughly $1.60 a conversation and about $2,400 a month at 1,500. The 1,500 is my estimate: a mid-market DTC store runs 2,000 to 10,000 tickets a month, and 1,500 is the conservative pre-purchase slice of a 6,000-ticket store.
Scenario | Monthly at ~1,500 conversations | Effective $/conversation | Notes |
Manual (in-house agents) | ~$2,400 | ~$1.60 | ~4 min at $0.40/min |
Tidio (Lyro) | from ~$300 plus usage | varies by meter | Plus tier plus Lyro conversation usage |
Rep AI | ~$368 at 50K sessions | metered on traffic | Session meter, so model it off visitors |
My AskAI | ~$499 | ~$0.33 | Scale covers 2,000 tickets, then $0.10 each, plus $0.02 a task step |
Alhena AI | ~$1,359 | ~$0.91 | Scale $999 annual covers 1,200, then $1.20 a credit |
Intercom Fin | ~$1,485 | $0.99 | Per outcome, plus $19 a seat on the Intercom helpdesk |
Gorgias AI Agent | ~$550 plus ~$1,965 overage | ~$1.68 | Pro bundles 190 automated interactions, then $1.50, plus the ticket fee |
Yuma AI | not published | not published | Sales-gated; no figures on the site today |
Against Gorgias and Fin those are AI charges compared with AI charges: the helpdesk seats sit on both sides and cancel out.
Baymard's aggregation of 50 studies puts the documented cart-abandonment rate at 70.22%, and two of the top checkout-stage reasons are questions (delivery too slow at 20%, and not seeing the total cost upfront at 12%). Those are measured at checkout, so they undercount everything a shopper gave up on before the cart.
With session conversion at 2.03% in June 2026, a handful of recovered sessions a day is worth modeling against your own AOV before you pick a tool on price alone. Gorgias reports shoppers who converse with AI or an agent converting at 154% higher rates, but that's their own platform data with no control group, and conversation-starters self-select as high intent, so take it as a direction of travel.
Three of these vendors charge per resolution, outcome or interaction, so the bill climbs as the agent gets better at its job, hardest in the week you can least afford it. It's why we price per ticket, and I've written up that math in more detail. If you want to model your own numbers against a Gorgias bill, there's an ROI calculator that does it for you.
So which AI chatbot for ecommerce is best for pre-sales?
TL;DR: Rep AI takes the scoreboard at 62 of 90 for pure in-widget conversion, with Gorgias the pick if you're already a Gorgias store on Shopify, Fin if you're on Intercom, Alhena if you're off Shopify, and My AskAI if you want one agent covering pre- and post-purchase on any platform at a fraction of the per-conversation cost.
If your question is which of these does the most inside the chat window, it's Rep AI or Gorgias. Rep AI takes the top line at 62 because the full conversion path arrives built (catalog search, guided narrowing, add-to-cart, checkout, live Shopify codes) with attribution over the top. What you accept for it is a meter that charges for footfall and a security page that doesn't match its trust portal yet.
Gorgias, one point behind on capability at 44 of 60, has the better promo engine and the better in-chat product page, and asks you to already be a Gorgias store on Shopify and to accept two charges on the same conversation. Intercom Fin at 61 is the same call for anyone already living in Intercom, and Alhena at 57 is the one to shortlist if you sell on Magento, BigCommerce or Salesforce Commerce Cloud, since nobody else documents those.
Ours is a different case. On the six capability rows we're fifth of seven at 34 of 60, behind Rep AI, Fin, Gorgias and Alhena, and the in-chat conversion action is something you build rather than switch on.
What we do is one agent covering the pre-purchase question and the returns ticket that follows it, on any platform, inside the helpdesk you already run, at a flat $0.10 a ticket while the others meter conversations, outcomes or sessions. Buy one of theirs if you're buying a shopping assistant. Ours is the argument if you want an agent that answers the sizing question accurately and is still there at 2am three weeks later when the parcel goes missing.
However you start, don't let it recommend anything on day one. Put the agent in Internal Notes mode on pre-purchase questions for two to four weeks and read what it drafts: sizing and stock are where the wrong answers cluster, and you want to see them before a shopper does.
That's how our own customers did it: Edel Optics ran notes mode before switching to direct replies, and YouGarden ran a month of Freshdesk notes first. When something reads wrong, ask Echo why the agent answered that way and which source it used, fix the source, and move on.
If you want to try it on your own catalog, our Shopify integration page shows how the connector and the sales-agent Task fit together, and the free trial runs 30 days with every feature unlocked, unlimited tickets and no card.
FAQs
What is an AI shopping assistant?
An AI shopping assistant is a catalog-first AI agent built to convert. It reads your product catalog, narrows a shopper down to a pick, and takes a commercial action inside the chat: a product card, an add-to-cart, a checkout link, a discount code. Its metric is conversion rate and attributed revenue, which is what separates it from an AI support agent like ours: knowledge-first, and measured on resolution.
What is the difference between an AI shopping assistant and an AI support agent?
The metric and the data source. A shopping assistant reads the catalog and is measured on conversion; a support agent reads the knowledge base plus live order data and is measured on resolution. Buying one of each means two subscriptions and two knowledge bases that drift apart, which is why Gorgias and Intercom Fin merged them into one agent with skillsets or roles while Yuma still sells Sales AI and Support AI separately.
What is the best AI chatbot for ecommerce?
On the nine criteria in this post, Rep AI scores 62 of 90, Intercom Fin 61, My AskAI 59, Alhena 57, Gorgias 53, Tidio 50 and Yuma 32. Rep AI takes it on the pre-built conversion path, and Fin is the pick for anyone already on Intercom. My AskAI's third place comes from setup, cost and platform coverage; on capability count we sit fifth of seven, so the right answer depends on which of those axes your store cares about.
What is the best AI customer support tool for ecommerce?
Rep AI, Intercom Fin and My AskAI lead the pre-purchase half of this scoreboard, in that order. If the queue hurting you is the post-order one (tracking, returns, refunds, subscriptions) then Yuma is the deepest tool in this set, and it sits bottom of this scoreboard only because pre-sales isn't the job it was built for. Our wider roundup of AI customer service tools for ecommerce scores that half properly.
What is the best AI chatbot for Shopify customer support?
Rep AI tops this scoreboard on Shopify, and Gorgias AI Agent has the deepest in-chat selling of the set as long as you also run Gorgias as your helpdesk. Every tool here works on Shopify in some form. Gorgias's Shopping Assistant states it "Can be used with Shopify stores connected to Gorgias" and Yuma's Sales AI runs "on your Shopify store", so both are Shopify-only on the pre-sales half, and Tidio's add-to-cart covers Shopify and WooCommerce. Fin is native on Shopify with a catalog connector elsewhere, Alhena has the broadest documented platform list, and we run through your helpdesk on any platform with Shopify as the pre-built connector.
What is the best AI customer service tool for D2C brands?
For a D2C brand whose chat is mostly browsing questions, Rep AI and Alhena are built for that job, and Alhena publishes Tatcha and Victoria Beckham as proof. Where one small team answers the sizing question and the WISMO ticket that follows it, one agent across both halves beats two subscriptions, which is the pattern at Edel Optics, YouGarden and Swytch on a flat per-ticket bill. Tidio is the cheapest way in for a smaller Shopify store, and Burker is the D2C brand it publishes.
Is there a free AI shopping assistant?
Three of the seven run a real free tier: Alhena, Tidio and Rep AI. Everyone else is on a trial, and ours is the longest at 30 days with every feature unlocked and no card.
Vendor | Free tier | Trial |
Rep AI | 100 visitors a month | 14 days |
Intercom Fin | None | 14 days, unlimited outcomes, no card |
My AskAI | None | 30 days, every feature, unlimited tickets |
Alhena AI | 25 conversations a month | Not published |
Gorgias AI Agent | None, Starter at $40 a month | Not published |
Tidio | 50 billable plus 50 Lyro conversations | 7 days |
Yuma AI | None | 30 days on your real ticket data |
What is agentic commerce?
Agentic commerce is an AI agent taking commercial actions for the shopper inside the conversation: recommending, adding to cart, applying a discount code, moving to checkout. It's real in four places here: Gorgias's in-chat product page with variant selection and add-to-cart, Alhena's agentic checkout, Rep AI's add-to-cart plus in-chat discount apply, and Fin's Get Cart and Update Cart connectors. Elsewhere the label covers a recommender and nothing more.
