ChatGPT Shopping Insights and Ads Radar for Sponsored AI Placements

ChatGPT can name a brand in prose, show a shopping unit, or show a sponsored placement. Those are three rows. Mixing them is how ecommerce QBRs go wrong, because a name in prose is an organic mention, a shopping unit is a product surface, and a sponsored placement is a paid unit. They move for different reasons and they get fixed by different teams, and a slide that folds them into one number is a slide no one can act on. We turn on Promptwatch ChatGPT Shopping on Professional or Business after the prompt list is stable, not instead of it. Ads Radar stays on its own row.

We do not publish Shopping conversion rates we did not measure. Visitor analytics (script or GTM) is the only place we will call a session. A conversion rate we did not measure is a number we cannot defend, and an undefendable number in a QBR is the number that gets the whole report questioned.

Shopping insights

ChatGPT Shopping tracking covers Visibility Score, sentiment, shopping-specific data, and an entity tracker. Shopping Insights shows the share of responses that include shopping features and whether your own products appear. Product recommendation tracking is "who got named in a shopping context," which is not the same as a citation on a how-to prompt. A shopping context is a different surface with a different intent, and the report that treats it like a how-to citation is the report that misreads the signal.

We load the products the PIM already sells. We do not let an agent invent SKUs. Content Agents can draft a category explainer into Webflow or Framer and wait in the review inbox. Specs come from the catalog or they do not ship. WordPress publish is still coming soon. The discipline here is simple: the catalog is the source of truth for what the brand sells, and an agent that invents a SKU is an agent that invents a product, and an invented product in a live page is a support ticket waiting to happen.

Professional at $245/mo lists shopping insights on the brand side. Business is $579/mo. Essential at $95/mo is the mention-and-citation program we start on, but it does not include Shopping Insights. Explore is free ChatGPT, 10 prompts, and is not a shopping stack. An agency project still needs a contract that lists Shopping; we do not infer it from unlimited projects. G2: 4.7/5.

Paid engines remain ChatGPT, Gemini, Claude, Perplexity, Grok, Llama, DeepSeek, Mistral, Copilot, AI Overviews, and AI Mode. Shopping is a ChatGPT surface. We do not pretend Gemini has the same unit. A shopping unit on ChatGPT is a ChatGPT fact, and generalizing it to engines that do not have the unit is the kind of thing a sharp client catches.

Ads Radar

Ads Radar tracks sponsored placements in AI search: which competitors bid on your prompts, when they appear, ad copy, claims, and positioning. That is paid competitive intel. It is not organic GEO. The split matters because the fix is different. An organic miss is a content and crawl problem. A sponsored unit is a media problem, and the team that owns one does not own the other.

We put Radar on a separate slide. If a competitor bought the unit, the fix might be a media conversation, not a rewrite of the FAQ. If nobody is buying and you are still absent from the shopping module, that is a product-feed and entity problem, not an H1 problem. The diagnosis drives the work, and a slide that mixes paid and organic drives the wrong work.

Otterly at $29 (Gemini add-on, lag) will not give you this split. Peec at $95 is mentions. Profound Starter at $99 annual is ChatGPT mentions without this commerce pair. We will not invent a BrightLocal or GeoRanker shopping SKU to fill the cell. A cell filled with an invented SKU is a cell that breaks the first time someone clicks through to the vendor.

Daily paid checks. No instant alerts. No $29 Promptwatch plan.

What we will not do

We will not call a sponsored unit a citation. We will not promise that appearing in Shopping equals revenue. We will not turn Radar on in week one while aliases are still wrong. We will not autopublish a comparison page that invents a competitor price we saw in an ad. Each of these is a specific failure mode we have seen, and the rule is the failure mode written down so it does not repeat.

FAQ

Do we need Shopping if we are B2B and have no cart?

Usually no. Keep prompt tracking and citations. Turn Shopping on if ChatGPT already shows a product module on your category prompts. If it never does, Radar may still matter if someone is buying the query.

Can we see Shopping in Explore?

Explore is 10 ChatGPT prompts for a demo. It is not the shopping insights SKU we run on a catalog. Use Professional or Business when Shopping Insights is required. Essential does not include it.

How we run it

1. Freeze prompts and aliases. Baseline mention vs citation for two weeks.
2. On Professional or Business, turn on Shopping Insights for the SKUs the PIM lists. Turn on Ads Radar on the same prompt set when that view is in scope.
3. Split the monthly slide: organic mention, shopping module, sponsored unit.
4. Measure clicks in visitor analytics if a session arrives. Do not invent a CVR.
5. hello@1001seomedia.com if you want the commerce views on the retainer. Bring the product list. Do not bring a promised Shopping rank.