AI Shopping: Correcting Data to Avoid Losing Sales

Key Takeaways

1AI responses are evolving rapidly, making one-time audits quickly obsolete

2Adobe forecasts a +130% increase in AI traffic on U.S. retail sites during the holidays

3Three types of errors: absence, incorrect data, purchase links captured by third parties

4Methods exist to audit, correct, and monitor flows and pages

Why it matters : Merchants who control their product data and monitor their AI visibility maximize their chances of capturing traffic and sales during seasonal peaks.

Product cards generated by ChatGPT and Google can display incorrect prices, promote the wrong variant, or direct the buyer to a reseller. With AI traffic expected to surge during the holiday season, there are methods to audit your key queries, correct your feeds, and secure the purchase click on your pages.

Unstable results require daily monitoring

AI-generated responses are continuously evolving, making one-time checks quickly obsolete. Repeating the same checks and logging errors allows for tracking variations by platform, region, and day, with a particular focus on daily checks. The audit should also be conducted by category, as a brand may be highly visible in headphones but absent in speakers. To automate these tasks, Writesonic's AI Shopping Tracker performs daily checks on ChatGPT and Google AI Overviews and indicates how often your queries trigger shopping results. Quick wins are recommended before Black Friday.

Recent visibility figures reveal important insights

Adobe forecasts a 130% increase in AI-related traffic on U.S. retail sites during the holiday season. In a single day in July 2026, the share of products chosen by ChatGPT from integrated feeds jumped from 8.26% to 61.54%. Among 517 stores that recorded visibility fluctuations of at least 33%, 450 experienced a decrease. In June 2026, a survey of over 2,000 brands revealed that when a brand is mentioned on YouTube or Reddit, it is about three times more likely to appear in AI-generated responses.

Where ChatGPT and Google product cards come from

In March 2026, OpenAI expanded its Agentic Commerce Protocol to product discovery, allowing merchants to share their feeds to display their catalogs in ChatGPT. Retailers like Target, Best Buy, and Wayfair are already integrated, and Shopify merchants are using Shopify Catalog. OpenAI introduced Instant Checkout in March 2026, allowing payments on merchant sites, with ChatGPT stepping in to attract the buyer before conversion on the product page. On Google's side, AI Overviews display product cards for many searches, relying on the Shopping Graph. Merchants feed this graph by sending feeds via Google Merchant Center. Google claims that the Shopping Graph contains over 50 billion listings. Data sources for ChatGPT include ACP feeds, Shopify Catalog, third-party providers, and public reviews, while for Google, it includes the Shopping Graph and Merchant Center.

Le brief IA que lisent les pros

Le brief IA que les pros lisent chaque soir

Les 7 actus IA du jour, décryptées en 5 min. Gratuit.

Inclus dès l'inscription : notre sélection des meilleurs guides & comparatifs IA.

Choisissez votre rythme

Gratuit · Pas de spam · Désabonnement en 1 clic

The three categories of errors that cost sales

The observed issues fall into three categories. First, absent or relegated products, such as a best-selling jacket that does not appear in a query under $300, or a model ranked fourth behind three competitors. Second, incorrect data on prices, variants, or specifications: a card might show $129 instead of $99 for Black Friday, recommend a running shoe for a trail query, or indicate a battery life of eight hours instead of fourteen. Finally, the purchase link may point to a reseller rather than the brand's site; if Zappos often receives this link, it may indicate an investment focus. Most brands discover these cases by chance, correcting one product while many queries remain unchecked.

Diagnosing with realistic queries and prioritizing by revenue

Diagnosis begins with constructing purchase queries that closely resemble actual usage, rather than simple product names. This involves assembling queries by category, use case, budget, and gift scenarios, then noting the content of product cards for each: items, prices, reviews, justifications, and links. These tests should be executed on ChatGPT, AI Overviews, AI Mode, and Gemini in offline sessions, comparing results to competitors to identify areas of deficiency, and then prioritizing corrections based on revenue impact, starting with best-sellers and high-margin categories.

Aligning prices and stock across all channels

Errors stem from a lack of consistency between feeds, product pages, and ads among retailers. When these sources diverge, the AI must choose and may make mistakes. The first focus should be aligning prices and stock across all platforms. The second focus is clarifying identity with a product schema and identifiers such as GTIN, MPN, brand, and variants to avoid confusion between models, sizes, or colors. The third focus is updating use cases and review information, which systems heavily rely on to justify their choices.