// sector · ecommerce
Before opening a single store tab, a shopper tells ChatGPT what they want to do with the product, what limits them and what they intend to spend. Your product slips into that reasoning, and the count is made in orders.
I cycle 12 km to work with two steep hills, which e-bike for under €2,000?
With hills every day, motor torque counts for more than the advertised range. Look for a mid-drive motor rather than a hub motor, a removable battery you can charge at the office and hydraulic brakes. At that price, be wary of listings that don't state the total weight.
Commuter e-bike
Mid-drive motor, hill-tested range, ships built.
your-store.com
// why this sector
Above €100 per basket, people rarely buy on impulse. The intended use and the price ceiling get talked through with ChatGPT first, which receives the whole equation where a Shopping query only catches a piece of it.
Twelve kilometres, two hills, an 18 m² living room with a cat: these real-life details settle the product long before a brand name does. An ad that picks up the constraint (torque on a climb, a fabric that washes) speaks to someone with no favourite yet, and who therefore stays open to changing their mind on brand and price alike.
Order matters. ChatGPT starts by explaining what to look at before naming a single product, and an ad that carries one of those criteria lands right when the shopper is learning how to choose, which is before any preference has set.
Above €100 per order, an entry CPC of $3 to $5 leaves room to learn. Below that, you would need a conversion rate nobody can promise. So we run the numbers before launching.
// intents
Typical prompt
"Which camera for travelling light and beating a phone, without becoming an expert?"
What the ad has to do
Tie a product to a precise use: the ad names the situation the shopper is living, never the spec sheet.
Typical prompt
"A mattress for a couple, one side sleeper, one who runs hot, under €800"
What the ad has to do
Hold the price constraint and justify it, with a single argument and the price in plain sight on the landing page.
Typical prompt
"Between these two strollers, which one for city life with no car and a lot of subway?"
What the ad has to do
Lead with a difference the shopper can check themselves, the comparison already laid out on the landing page.
Typical prompt
"An alternative to the best-known cordless vacuum for a flat with two cats"
What the ad has to do
Stand next to the category reference without quoting it in the ad, then lead to the category page filtered on the case.
// context hints
For a store, a hint describes the buyer's constraint more than the product: the size of the balcony, the tight budget, the cats. Three use-case families, taken from an equipment store:
Generate your context hintsIllustrative excerpt. The real structure is built from your search terms, sales calls and support tickets.
AG · Precise use case
AG · Tight budget
AG · Comparison and gifts
// scenarios
Three examples that tie a conversation to an ad, a page and a metric. They are not delivery promises.
// measurement
OpenAI's native report lists impressions, clicks, spend, CTR, average CPC and average CPM. Not a single order. We connect the OpenAI Pixel and the Conversions API to your purchase events, and the account is then read against the Hiints Index.
order_created
The primary event. Without it the channel runs on CPC, which says nothing about your profitability.
checkout_started
The intermediate signal. The gap with orders shows whether the fault lies with the ad or the page, which saves you rewriting one when the other is what's stuck.
Attributed order value
Set it against the site-wide average, because if the channel brings smaller baskets, the target CPA moves with it.
Margin per order
The real unit of decision. Ad groups are arbitrated on the margin they bring in, never on their CTR.
CPA vs your other channels
Same basket, same period, with Shopping and Meta alongside: no other comparison justifies moving budget from one channel to another.
Max CPC vs Hiints Index
Your margin per order sets the CPC you can afford. The Hiints Index shows where comparable ecommerce accounts sit, while OpenAI publishes no reference point.
// honest qualification
Yes, clearly
Your average order value is above €100 and your products are chosen on criteria: use, size, compatibility, budget. Your category or product pages are clean, and your purchase events are already tracked, or can be before the first euro is spent.
Yes, with one caveat
Your basket sits close to €100, or your product is bought without much comparing, or your catalogue is wide while your category pages don't yet mirror the use cases. You sell in a category covered by OpenAI's advertising policies (supplements, financial products): eligibility is checked before any launch.
Not yet
If your monthly media tops out at €3,000, or the margin on an order can't absorb an entry CPC of $3 to $5, the test won't teach you anything useful. You may as well keep your current channels. We'll tell you so without detour.
Free setup, done in under 48h. Then 15% of ad spend under €5,000 of monthly media, 10% above. See pricing →
// questions
Yes, when the product is chosen on criteria and the average order is above €100. A €15 consumable bought without thinking leaves no room between the CPC and the margin. We run that calculation with you before you sign.
No. The shopper described a precise use and the landing page has to carry it on: product page if the intent is clear, filtered category page if they're still hesitating. The homepage restarts the conversation from zero.
Purchase events: checkout_started, order_created, then attributed order value and margin. The native report stops at the click, which is why the OpenAI Pixel and the Conversions API cover everything after it, down to a margin that can be compared, line by line, with the CPA of your other channels.
Not as a first step. We start with the use-case families where margin and pages hold up. When the product feed allows it, a catalogue campaign (a distinct format) comes next, judged on margin like everything else.
No. Shopping captures a named product search, Meta creates desire, ChatGPT receives the need stated with its criteria. The three get compared on CPA and margin per order, at equal basket size.
The budget follows from the margin on an order: the wider it is, the more orders an entry CPC of $3 to $5 leaves you to win before turning profitable. Technically the floor is $25 a day per campaign, but under €3,000 of monthly media we advise against testing. Setup is free; the fee runs at 15% while media stays under €5,000/month, then 10%.
// keep reading
How Much Does ChatGPT Ads Cost in 2026: CPC, CPM and Minimum Budget
Observed CPC of $3 to $5, CPM of $25 to $60, $25 per day minimum per campaign, no overall minimum since May 2026. What sets each cost line.
Read →ConversionChatGPT Ads Landing Pages: The Page That Continues the Conversation
The landing page is one of OpenAI’s ad selection signals. One page per family of intent, the checklist, three prompt → page pairs and message match.
Read →TrackingMeasuring ChatGPT Ads Conversions: Pixel, API, UTM and CRM
The native report has six columns and no conversions. OpenAI Pixel, Conversions API, a UTM convention, events by business model and CRM reconciliation.
Read →// shall we talk?
Thirty minutes. We set your average basket and your margin beside what the Hiints Index shows for your category, and if the numbers don't hold, we'll say so.
Book a call30 minutes · no commitment · zero pitch
// other sectors
