Make every product page quotable
Product schema with price, stock, materials, fit and rating, in HTML rather than JavaScript. This is the difference between being priced into an answer and being invisible.
Optimise Content →Solutions · By industry
Shoppers ask AI first. It names three products.
Shoppers ask AI for the best product before they browse, and the answer names a few brands with prices attached. Prefer tracks every shopping prompt in your category, shows which reviews and roundups decide them, and turns your product pages into ones AI will quote.
“best sustainable running shoes”Category prompt
Per engine
ChatGPTAbsent
PerplexityAbsent
ClaudeAbsent
ChatGPTAbsentPerplexityAbsentAI OverviewsNamed
ClaudeAbsentNamed in the answer
Allbirds
On
VejaWho decides it
wirecutter.comOne editorial roundup supplies most shopping answers
Shopping answers name three to five products with prices. Absence here is the whole funnel, because the buyer never browses.
The shelf math
A category of hundreds. An answer of three.
How AI engines answer shopping questions, across 380 tracked retail domains.
68%of shopping answers name 5 or fewer products
The consideration set is built before the buyer browses.
5.2×editorial roundups and reviews out-cite brand PDPs
Wirecutter, Reddit and Trustpilot carry the recommendation.
51%of D2C brands have no price an engine can read
No Product schema means no place in a price prompt.
1-4 wksfrom shipped fix to measurable movement
Faster than B2B, because retail sources re-crawl often.
Prefer benchmark · 380 D2C and retail domains · 6 engines · Q2 2026.
The shortlist · 01
The shortlist is built before anyone browses.
Shoppers used to open five tabs. Now they ask once and get three products with prices and a verdict. Miss that answer and you are out of the comparison, and no analytics tool will show you the visit you lost.
“best sustainable running shoes”
ChatGPT, answered this morning.
Three stand out. Allbirds Tree Dasher 2, around $135, uses eucalyptus fibre and sugarcane foam and is the most comfortable for daily wear. On Cloudrunner, about $150, offers better support for longer distances. Veja Condor 3, roughly $160, is the most transparent on supply chain but runs narrow.
Sourceswirecutter.comallbirds.com
reddit.com
3products named3prices quoted0times Fernwell appears
Who decides · 02
Reviewers and roundups pick the products.
Retail citations go to editorial reviews, community threads and review platforms. Brand pages get cited to confirm price and specs, after the recommendation is already made.
Where citations come from
One tracked category, 260 prompts, last 30 days.
- Editorial roundups
- 31%
- Review platforms
- 19%
- Community
- 18%
- Retailer & marketplace
- 14%
- Competitor-owned
- 10%
- Your owned pages
- 8%
DomainTypeCitation share
1wirecutter.comNot included in 2 relevant guidesEditorial14.6%
2reddit.comr/RunningShoeGeeks, sizing threadCommunity11.3%
3
trustpilot.com4.1 stars, sizing theme recurringReviews8.9%
4
runnersworld.comEditorial7.8%
5
amazon.comListing lacks materials attributesMarketplace6.4%
amazon.comListing lacks materials attributesMarketplace6.4%6allbirds.comTheir PDPs are fully structuredCompetitor5.7%
7
rei.comRetailer4.9%
8
fernwell.comPDPs missing Product schemaYour domain3.4%
Domains cited across 260 tracked shopping prompts, last 30 days. Editorial roundups decide the recommendation; your pages confirm the price.
The prompt map · 03
Six families decide a purchase.
Two of them do not exist in software: price and occasion. Both move fast, both are seasonal, and both are won with data rather than copy.
Fernwell, per family per engine
Status of the best prompt in each family.
CitedNamedAbsent
FamilyExample promptChatGPTPerplexityAIOClaudeWhat wins it
Categorybest sustainable running shoesEditorial roundups + a structured PDP
ComparisonAllbirds vs FernwellYour own honest comparison pages
Attributewaterproof trail shoes, wide toe boxSpec attributes as structured data
Pricebest running shoes under $120Product schema with a live offer price
Reviewsare Fernwell shoes worth itReview volume + the complaint theme fixed
Occasiongifts for a runner under $150Gift guide placement, pitched a quarter early
Live from the Fernwell demo workspace. Occasion prompts are tracked year-round, because Q4 answers are decided in Q3.
The playbook · 04
Six plays win retail categories.
Each play maps to the Prefer tool that runs it, so this is a queue, not a PDF.
Win the editorial roundups
Wirecutter, Runner's World and category guides out-cite brand pages five to one. Prefer finds the guides you are missing from and briefs the pitch.
Action Center →Answer attribute questions
Prompts about fit, material and use case are where a small brand beats a giant. They need spec pages AI can read, not lifestyle copy.
Create Content →Fix the repeated complaint
One recurring theme, sizing, shipping or returns, becomes the warning attached to your name in every answer. Prefer traces it and tracks the repair.
Answer Engine Insights →Publish gift guides early
Seasonal answers are decided a quarter ahead, when AI crawls the guides. Ship in Q3 to win Q4.
Create Content →Watch the shelf move
Every product named in your category, per prompt, per engine, per week. When a rival enters an answer you owned, you hear that day.
Competition →D2C vs marketplace · 05
Own store and marketplace lose differently.
A direct brand fights to be named at all. A marketplace seller fights to be the listing AI picks. Prefer ships a prompt set for each.
Direct-to-consumer
You own the product page and the reviews, so the job is making them quotable. Where you are exposed:
Category“best X for Y, most sustainable X”
Attribute“wide fit, waterproof, vegan”
Reviews“is X worth it, X sizing”
Occasion“gifts for X under $150”
Exposure: a competitor's structured PDP gets priced into the answer, yours does not.
Marketplace seller
The listing is the page AI reads, and the platform controls it. Where you are exposed:
Price“cheapest X, best X under $50”
Availability“in stock, ships by, next day”
Listing quality“attributes, images, Q&A”
Seller trust“X seller reviews, is X legit”
Exposure: an incomplete listing means the marketplace recommends someone else's.
Proof · 06
Better pages. Named in answers.
“We were spending on paid social to fix a problem that was really a markup problem. Once every PDP had real product data and we landed two roundups, assistants started recommending us with the price attached.”
6% → 34%mention share on category prompts0 → 71%PDPs with valid Product schema4 weeksto first measurable move+28%assisted sessions from AI referrals
Questions
Asked plainly.
How do ecommerce brands get recommended by AI assistants?
By being readable and reviewed. Engines build shopping answers from editorial roundups, review platforms and community threads, then confirm price and specs from product pages that carry structured data. A brand with no Product schema cannot be priced into an answer at all.
What is AEO for ecommerce?
Answer engine optimization for retail means earning a place in AI shopping answers rather than a rank in a results page. It combines structured product data, editorial and review placement, and fixing the complaint themes engines repeat about your brand.
Which shopping prompts should a D2C brand track?
Six families: category (best X for Y), comparison, attribute (fit, material, use case), price, reviews, and occasion or gifting. Price and occasion are unique to retail, both are seasonal, and both are won with data rather than copy.
Do AI assistants use my product feed?
Google AI Overviews draws on Merchant Center data, and every engine reads Product schema on your PDPs. Keeping price, availability and attributes accurate in both is the baseline for appearing in price and availability prompts.
Which sources do AI engines cite for shopping answers?
Editorial roundups first, then review platforms and community threads. In Prefer's Q2 2026 benchmark of 380 retail domains, roundups took 31% of citations, review platforms 19%, community 18%, and brands' own pages 8%.
How do I get into AI gift guides?
Ship and pitch early. Seasonal answers are largely decided a quarter ahead, when engines crawl the guides being published. Prefer tracks occasion prompts year-round so the pitch happens in Q3, not December.
Can AI answers hurt a brand's reputation?
Yes, and quietly. Engines summarise your review corpus, so a recurring theme like sizing or shipping becomes a caveat attached to your name in every consideration answer. Prefer traces the theme to its source so it can be fixed where it lives.
How fast do retail AI answers change?
One to four weeks, faster than B2B, because retail sources are crawled frequently and prices change often. Structured data fixes usually show within the first crawl cycle.
Get your free AI visibility report
in about 10 minutes.
See how answer engines describe your brand today, and where the openings are to outpace the competition.