Oossa

AI shopping assistants drive modest product page visits but few direct clicks

Study links AI purchase suggestions to later browsing on 20 online retailers, finding a rise in product page visits but little referral traffic.

By Published by Oossa: 1 min read

Clay Banks · Unsplash

Researchers examined how people act after asking a voice or chat assistant what to buy. They matched the assistant's answer to the shopper's later activity on twenty large e‑commerce sites. When a shopper had not visited a store in the previous week, the chance of opening the product page named by the assistant jumped from 0.02 to 2.07 visits per 100 answers within a day. That is a 2‑point increase, with a confidence range of 1.7 to 2.4 points.

Even with the higher visit rate, most traffic did not come through a direct link from the assistant. Eight‑nine percent of the product‑page visits showed no trace of the assistant’s referral. Over half of the visits (54%) started with a web search, the same proportion as in the shoppers’ usual browsing patterns.

When a search happened just before the first product‑page visit, the query named the assistant’s suggested item 28 % of the time, compared with only 1 % in earlier weeks. This suggests that users treat the search box like an address bar after hearing a recommendation.

What the numbers mean for retailers and users

The authors propose a metric called “product follow‑through”: the share of AI answers that lead to a visit to the named product. It can be measured without needing referral data from the assistant provider, making it useful for evaluating shopping bots.

For retailers, the modest lift in product‑page visits shows that AI suggestions can steer traffic, but the lack of direct referrals means the assistant does not guarantee a sale. Users still rely on their own search habits, often typing the product name themselves.

The study covered two separate five‑month periods, and the pattern held across both, indicating the behavior is stable over time.

Why it matters

For everyday shoppers, an AI recommendation can make them more likely to look at a specific product, but they will still usually type their own search. Retailers can use the product follow‑through metric to gauge how well their assistants are nudging traffic, without needing data from the assistant provider.

Was this article useful?
Share

Read next

Oossa · Newsletter

The week in AI, explained

Every Monday: the stories worth knowing, in plain language. Free, no spam.