Worth reframing this before you chase the AI part: whether your offers "show up" for a shopping agent is a data problem, not a magic problem. An agent can only surface a reward it can actually read in a clean, structured place. Most brands lose here because their best offers live in a banner image, a hero graphic, or buried in PDP body copy — invisible to anything that isn't a human eye.
Three things I'd check, in order:
1. Get the offer out of marketing and into structured fields. Discount, eligibility, and expiry belong in your product feed and Offer schema (price, priceValidUntil, eligibleCustomerType), not in a JPG. If a machine can't parse it without OCR, assume it's missing.
2. Loyalty pricing is the hard case. Agents almost always see the public price, not the logged-in member price. If you want member rewards reflected, you either expose a member price through an API/feed the agent can hit, or you publish the promo as an everyone-eligible offer. Pick one on purpose.
3. Feed hygiene decides trust. If your site, your Google feed, and your marketplace listing disagree on price or availability, the agent drops you or shows the wrong number. One source of truth, reconciled daily.
Start by Worth reframing this before you chase the AI part: whether your offers "show up" for a shopping agent is a data problem, not a magic problem. An agent can only surface a reward it can actually read in a clean, structured place. Most brands lose here because their best offers live in a banner image, a hero graphic, or buried in PDP body copy — invisible to anything that isn't a human eye.
Three things I'd check, in order:
1. Get the offer out of marketing and into structured fields. Discount, eligibility, and expiry belong in your product feed and Offer schema (price, priceValidUntil, eligibleCustomerType), not in a JPG. If a machine can't parse it without OCR, assume it's missing.
2. Loyalty pricing is the hard case. Agents almost always see the public price, not the logged-in member price. If you want member rewards reflected, you either expose a member price through an API/feed the agent can hit, or you publish the promo as an everyone-eligible offer. Pick one on purpose.
3. Feed hygiene decides trust. If your site, your Google feed, and your marketplace listing disagree on price or availability, the agent drops you or shows the wrong number. One source of truth, reconciled daily.
Start by listing every active promo and asking one question per promo: could a machine read this right now, yes or no? That audit usually finds the leak fast.
Happy to walk through an offer/feed audit with you if it's useful. Listing every active promo and asking one question per promo: could a machine read this right now, yes or no? That audit usually finds the leak fast.
Happy to walk through an offer/feed audit with you if it's useful.
Interesting challenge, a few questions before diving in:
Are you trying to surface these offers within your own storefront, or across third-party channels? And are users currently seeing the rewards but not engaging, or are they not seeing them at all?
The problem usually sits in one of two places,either the offer is showing at the wrong moment in the journey, or the personalization layer isn't connecting what the user did to what they're being shown.
Brands that solve this well typically focus on understanding why users aren't engaging with the offers in the first place, whether it's visibility, timing, relevance, or trust. Without that diagnostic layer, most loyalty programs end up optimizing the wrong variable.
The agentic layer is only as good as the data and insights feeding it, that's usually where the gap is.
Happy to jump on a call and dig into your specific setup, would be useful to understand your current stack and where the rewards are dropping off.
If you're looking for brands to talk to, the ones actually doing this today are mid-market DTC teams with a headless stack and an in-house data person — not enterprise. That's your ICP for outreach. The tell is a brand that already maintains a clean product feed for Google Shopping and a loyalty program with an API (Yotpo, Smile, LoyaltyLion). Those two facts mean they can technically expose member offers to an agent; everyone else is still stuck in banner images.
From the go-to-market side I'd stop pitching "agentic commerce" and pitch the measurable gap: their public price is what agents see, so members get quoted the non-member price and the agent recommends a competitor. That's a revenue leak a growth lead will meet about; the AI framing is a conference talk. Ask for their feed URL and show them one product where the offer is invisible — that's a 15-minute discovery call that closes itself.
Next step: pull 10 brands off BuiltWith filtered for a headless front end plus a loyalty app, and open with that one-product audit. Happy to go deeper on the outreach sequence on a call.
I've been in the automation / AI-agents lane more than pure e-com, but the pattern is the same: agents only surface what machines can read.
Before you chase a fancy shopping agent, audit each active promo with one question: can a system read discount, eligibility, and expiry without looking at a banner image? If the answer is no, put the offer in structured fields / feed / API first. Loyalty pricing is the trap — agents usually see the public price unless you expose a member price on purpose.
On the ops side, I wire the boring half with n8n or Make: promo changes → feed update → Slack alert when price/availability disagree across channels. That hygiene is what makes agentic discovery trustworthy later.
If useful, I can sketch a 30-min audit of your offer → feed → alert loop: https://clarity.fm/charlesouzeau
