Should You Fix Product Data Before Spending on AI Visibility?

Should we fix our product data first, or start tracking AI visibility now? Ecommerce leads ask this in almost every first conversation, usually because AI visibility is "on the back burner but we need to focus on it" and the catalog has known gaps. The short answer: do a first pass on your top revenue products, then start tracking. Being mentioned by ChatGPT or Google AI Mode is only useful if what the assistant says about your product is true.

A survey commissioned by ACI Worldwide and published on June 29, 2026 puts a number on the cost of getting it wrong. Six in ten UK shoppers (60%) said they would stop using an AI shopping agent after one mistake, per a YouGov poll of 2,080 UK adults. One wrong price, one missing size, one false attribute, and the shopper is gone. That changes what "winning" in AI shopping means.

What the ACI Worldwide research found

The headline numbers describe a trust gap rather than a capability gap. In the ACI Worldwide and YouGov survey, 19% of UK consumers trust AI assistants to make everyday purchasing decisions, compared with 55% who trust a human expert or adviser. Seven in ten (69%) say they do not trust AI even when it follows rules they set themselves.

The most operationally important finding is the fragility. Sixty percent would abandon an agent after one mistake. That is a single strike threshold for a behavior most people have only recently started trying. And 54% believe the technology company that built the agent, not the retailer, should be accountable for refunds when something goes wrong, which tells you how unforgiving shoppers expect to be when it fails.

This sits on top of a market that is still early. Forrester's December 2025 Consumer Pulse Survey found that 23% of Gen X US online adults had used ChatGPT to search for products in the past month, rising to 35% of Gen Z. Adoption is real but early, which means every early experience carries outsized weight. A shopper who gets burned in month one does not come back in month two.

Why a wrong answer costs more than a missing one

When an assistant omits your product, you lose a chance. When it recommends your product with the wrong price, a missing size or an attribute that turns out to be false, you lose the shopper and you hand the platform a reason to distrust your data. Those are different failure modes with very different costs.

Here is the mechanism. Google describes AI Mode as using query fan-out, breaking one shopper question into subtopics and issuing many searches at once. "Best sunscreen for sensitive skin under $20" becomes narrower questions about mineral versus chemical formulas, fragrance-free options, SPF, price and reef safety. The answer is assembled from the product facts that match each one.

If your data answers most of those questions correctly and one of them wrong, the assistant may still recommend you. The shopper then discovers the error at the worst possible moment.

The failure modes, from least to most expensive:

• Omission. Your product never appears. One lost opportunity, recoverable.
• Stale price. The answer quotes a price that no longer exists on your site. Lost sale plus a trust hit on the assistant.
• Wrong attribute. A "fragrance-free" product that is not. Abandonment, a return, and possibly a shopper who stops using the channel.
• Thin data. A generic match with nothing to differentiate you. Crowded out by the brand with richer, correct attributes.

The ACI finding makes the bottom three rows existential rather than annoying. In a market where six in ten shoppers say they leave after one mistake, an inaccurate recommendation sourced from your catalog is a liability you handed to the platform.

Accuracy is a discovery decision, not a back office task

Most brands treat catalog accuracy as an operations problem owned by a merchandising or data team, separate from the marketing question of visibility. That split no longer holds. The data an assistant reads to decide whether to recommend you is the same data it reads to describe you, and the shopper experiences both as one moment.

The brand with richer, correct attributes does not only get matched more often. It gives the assistant fewer chances to guess, and guessing is where wrong answers come from. A complete, accurate record on the page and in the feed reduces the surface area for error. That is why the first pass on product data comes before spending on more mentions: more mentions of a product the assistant describes wrongly is more exposure to the one mistake shoppers say they will not forgive.

Being the low error source is also something almost no brand manages deliberately, because most teams only find out how they appear in ChatGPT, Google AI Mode and Perplexity when someone asks for it. Tracking how your products are described across AI assistants turns that into a routine check rather than an occasional surprise.

What we check first

When a brand asks us whether to fix product data or start tracking, this is the first pass we run. It is scoped to the products that carry the revenue, not the whole catalog.

1. Price and stock agree across page, feed and answer. Pull the top 20 products by revenue and confirm that the price and availability on the product page, in the Merchant Center feed and in what ChatGPT and Google AI Mode currently say about the product all match. A feed can be technically valid and still produce wrong answers downstream.
2. Attribute accuracy on the top 20. Verify the five attributes shoppers actually filter on in your category: size, material, compatibility, formulation, whatever matters. Every claim on the page should be true and stated in the words shoppers use. This is the heart of the product content work.
3. Merchant Center disapprovals and limited products. Open the diagnostics and clear disapprovals, missing GTINs and price mismatches on the revenue list. Assistants that draw on Google's shopping data inherit whatever the feed says, so feed fixes are also answer fixes.
4. Sync cadence matches price velocity. If you reprice daily, a weekly feed sync guarantees stale data in AI answers. Match the refresh rate to how fast the catalog actually changes.
5. Then track. Ask the assistants the top buying questions in your category, record whether each product is named and whether it is described correctly, and check again after the fixes ship. Named and accurate is the metric that protects the channel.

Frequently asked questions

Does this research apply outside the UK?

The ACI Worldwide survey covered UK adults, so treat the exact percentages as UK figures. The underlying pattern, low tolerance for AI purchasing errors, lines up with Forrester's US data showing early and cautious adoption.

Is visibility in AI shopping still worth pursuing?

Yes. You cannot win a recommendation you never appear in. The point is that visibility is necessary but not sufficient. Once you are present, the accuracy and completeness of how you are represented determines whether the shopper converts or stops trusting the channel.

What counts as a mistake that loses a shopper?

A wrong price, an out of stock item presented as available, a false or missing attribute, or a product that does not match the stated need. Any of these can sour a shopper on the assistant. The ACI survey suggests a single instance is enough for many people.

How is this different from traditional SEO?

Traditional SEO rewarded ranking a page. AI shopping rewards being the correct, retrievable answer to many narrower questions at once. A page can rank well and still feed an assistant the wrong attribute. The unit of trust shifted from the page to the product record.

Does the purchase happen inside the AI assistant?

Mostly no. OpenAI removed native checkout from ChatGPT in March 2026, as Forrester's analysis of the pullback describes, so assistants recommend and send shoppers to the merchant's site to buy. Your product data drives the recommendation. Your site closes the sale.

The first chapter of AI shopping was about whether the assistant can find you. The ACI data opens the next one: when it finds you, does it get you right? If your team does not have the time to run the first pass, a Paz expert finds the gaps in your product information, you approve the fixes, and the platform publishes them and tracks what changed in the answers. Talk to us.