Real Conversion Results From Shopify AI Chatbots (2026)

The four stores, and what actually happened

Every figure below comes from a published Kinect case study, re-verified against production data on 2026-07-28. A shopper counts as engaged when they sent the rep a message or clicked a product it recommended, within the 14 days before they purchased. Loading a page with the widget on it does not count.

The volumes behind those rates: Bloom absorbed 26,900 visitors in week one and logged 158 engaged orders. Loosey Goosey has taken more than 1,000 conversations and 116 engaged orders. VTM has run over 4,000 operator conversations against a $2,850 capital purchase, worth $280K+ in engaged-operator revenue across 93 orders.

So the answer is yes, with a caveat that most of this category will not put in writing.

Why that table is not the answer you wanted

Think about who opens a chat widget on a product page.

Someone comparing two jackets they are close to buying. Someone who needs to know whether the size runs small before they commit. Someone checking the return window because they have already decided and are managing the downside.

Now think about who does not. Someone who bounced in four seconds. Someone who landed from a paid social ad and is browsing. Someone on their eleventh visit who already knows the answer.

The engaged group was going to convert better than the unengaged group if the widget had never been installed. Intent causes engagement and intent causes purchase. Comparing the two groups measures the intent gap plus whatever the assistant contributed, with no way to separate them.

That counterfactual is not observable in your analytics. It has to be constructed.

What the rest of the category publishes

We read the live claims pages of four competitors in September 2026 and checked each published figure for a stated methodology.

The Envive figure is the useful one to sit with, because it is the same statistic as ours. 12.3% against 3.1% is a chatter versus non-chatter split, exactly like our 13.8% against 1.8%. Neither number tells you what the assistant caused.

Zipchat deserves credit for one line most of the category will not write. Their page says to treat any single published rate, including the ones on that page, as directional. That is the right instruction, and it sits underneath eight numbers that read as anything but directional.

The pattern is not that these vendors are lying. The numbers are usually real. The inference a reader draws from them is usually not.

How to measure it on your own store

A holdout, which performance marketers know as an incrementality test. Randomly withhold the assistant from a slice of visitors, then compare the two randomized populations rather than the two self-selected ones. Four things have to be right, and each one is a place we have seen this go wrong.

Randomize at the visitor, and keep them there

Assignment has to be stable across sessions and devices, as far as your identity resolution can carry it. Randomizing per session lets the same person land in both arms across a multi-visit purchase path, which is most considered purchases.

The subtle failure is worse than the obvious one. If you change the split, say from 50/50 to 90/10 because the results look good, a naive bucketing implementation re-stamps returning visitors into new cohorts. Your history is now a blend of people who saw the assistant, then did not, then did. We have hit exactly this. The experiment does not error. It quietly stops meaning anything.

Measure revenue per visitor, not conversion rate of engagers

This one is entirely about denominators. The treatment arm is everyone who could have used the assistant, including the large majority who ignored it. If you narrow to engagers you have thrown away the randomization and rebuilt the original selection effect inside your experiment.

Revenue per visitor across the whole arm also catches effects a conversion rate misses. An assistant that raises average order value while leaving conversion flat is working. An assistant that converts more people into smaller orders may not be. Both are invisible in a conversion-rate readout.

Size it before you run it, and accept the answer

Ecommerce conversion rates are low and order values are skewed, which is a bad combination for statistical power. Detecting a few percent difference in revenue per visitor takes far more traffic than most people assume, typically tens of thousands of visitors per arm.

Plenty of stores cannot detect the effect size they care about in a reasonable window. That is a real finding about the measurement, not a reason to run the test anyway and report whatever came out. A 90/10 split makes it worse, because the small arm sets your power.

Know your attribution ceiling and state it

Linking a conversation to an order means carrying an identifier from a chat session through to a purchase event. That chain breaks in ordinary ways. A shopper switches from phone to laptop, storage gets cleared, a privacy setting drops the identifier, the purchase arrives by webhook without the client-side context attached.

So there is a ceiling on the share of influenced orders you can attribute, and it is well under 100% on real storefronts. Anyone claiming complete AI revenue attribution is either not looking or not saying. Publish your join rate alongside your attributed revenue, because a number without its coverage is unreadable.

The number we would rather be judged on

Across live stores, brands running Kinect have seen 3–6% more revenue measured against their own baselines. It is a smaller number than anything in the competitor table above, and it is the one we lead with.

It is still not a randomized estimate. Seasonality, promotions, traffic mix, and every other change shipped that quarter sit inside a baseline comparison. We say "measured against their own baselines" rather than "caused by" for that reason. The full method is on how we measure.

As far as we can tell, no vendor in this category has published a completed holdout result for an on-site AI shopping assistant. Not us either, yet. The stores with enough traffic to be powered have not run long enough, and the ones that have run long enough do not have the traffic. When that changes we will publish the result, including if it comes in smaller than what the category advertises.

The case study where we deleted the numbers

A.L.C. is a contemporary fashion brand on Shopify running Kinect across a 5,000-piece catalog. Their case study carries no conversion or order-value statistics at all.

The engaged-order sample was seven orders. Engaged average order value came in at $489 against $521 for unengaged, on a sample far too small to carry a claim in either direction. So we pulled the numbers and left the story qualitative.

That is the test worth applying to any vendor case study you read. Ask what got removed.

What to ask any vendor, including us

• Is this number a comparison of engagers to non-engagers? If yes, it is a selection effect, and everyone in the room should say so out loud.
• What is the denominator? Revenue per visitor across the whole arm, or conversion rate among the people who engaged?
• Was assignment randomized, and was it stable across sessions? Ask specifically what happens to returning visitors when the split changes.
• What is your join rate? Attributed revenue without its attribution coverage is unreadable.
• What did you remove from this case study, and why?

Ask us the same five. The methodology is published at how we measure, the underlying numbers are at case studies, and if you would rather have the conversation directly, book a demo.