• AI visibility

Guide · AI search

How can a mattress brand win customers who ask AI which mattress to buy?

By being one of the few mattresses an AI assistant names when a shopper describes how they sleep and what they can spend, and by giving that assistant facts it can check: materials, firmness, trial and return terms, and reviews from real owners. Mattress shoppers face too many look-alike choices, so an answer that narrows the field carries weight. That weight also makes accuracy and honest claims more important than in most categories.

By Underneath Team Guide Updated

The short version

  1. AI is already part of mattress research. In a survey of 1,006 US mattress shoppers run in August 2026 by the mattress maker Amerisleep, 37% used AI tools in their search and 49% started on Google.
  2. Shoppers are overwhelmed: in the same survey, 33% had delayed or abandoned a mattress purchase because the options were too confusing, and price drove the decision for 73%.
  3. The market is shrinking, so every sale is contested. The International Sleep Products Association reports US mattress wholesale value fell to $9.25 billion in 2025, with unit shipments down 13.2%.
  4. Mattress shoppers research and read reviews. In Better Sleep Council research, consumers who researched a mattress were most interested in comparing prices (64%) and consumer reviews (58%).
  5. Ratings move AI picks. In a test of AI recommendations for skincare, another product people cannot judge before using, a well-known brand’s lead disappeared when a rival had less than a 0.1-star rating advantage.

Who buys mattresses, and what is a customer worth?

Adults replacing a mattress years late, spending around a thousand dollars, and adding bedding afterward.

The Amerisleep survey (opens in a new tab) is a mattress brand’s own research, so read it as one source, but it maps the journey in detail. Respondents waited an average of 5 years after realizing their mattress needed replacing. They set an average spending ceiling of $982 for the mattress itself, from $606 for Gen Z to $1,109 for Gen X. After buying, 63% kept spending on toppers, pillows and bedding, about $74 on average. Memory foam led at 37% of purchases or planned purchases, followed by hybrids at 29%.

The industry is under pressure. The International Sleep Products Association (opens in a new tab) (ISPA) found total US mattress market value declined 6.5% in 2025 while unit shipments fell 13.2%; BedTimes (opens in a new tab), ISPA’s magazine, puts the 2025 total at $9.25 billion in wholesale value, down from $9.9 billion. The largest company in the category feels it too. Somnigroup (opens in a new tab), owner of Tempur-Pedic, Sealy, Stearns & Foster and Mattress Firm, reported second quarter 2026 net sales of $1,823.5 million, down 3.0%, with Mattress Firm at $922.2 million, down 2.8%, mostly from store closures. In a market this flat, growth comes from taking customers from competitors.

A sale is only worth something if the mattress stays. Online brands built the category on long trials: Casper promised a 100-night “risk-free” trial, and NPR reported (opens in a new tab) in 2020 that its “refunds, returns and discounts” were worth almost $81 million in 2018. A shopper who picks the wrong mattress costs the brand twice.

Where does AI already sit in the mattress buying journey?

At the research step, where shoppers compare types, brands and prices before choosing where to buy.

Mattress shoppers do their homework. In consumer research for the Better Sleep Council, ISPA’s consumer education arm, most consumers said they research (opens in a new tab) before buying a mattress, a larger share than in 2020. They were most interested in comparing prices (64%), consumer reviews (58%) and promotions (55%), and 48% wanted to compare product features. An earlier Better Sleep Council survey found shoppers used three to four information sources (opens in a new tab) on average, and almost 75% shopped both online and in store.

AI tools now sit among those sources. In the Amerisleep survey, AI use was fairly even across generations, from 32% of baby boomers to 39% of Gen Z. Across all US retail, AI visits are worth more than they were. Adobe data reported by TechCrunch (opens in a new tab) shows AI traffic to retailers converted 42% better than other traffic in March 2026, after converting worse a year earlier. Adobe does not break out mattresses.

The appeal is obvious. A shopper facing dozens of near-identical boxed mattresses wants someone to narrow the field. A third of shoppers in the Amerisleep survey gave up or delayed because the options were too confusing. An AI answer that names three options for a side sleeper on a budget solves that problem, for better or worse.

Which questions lead shoppers to a mattress brand?

Questions about how they sleep, what they can spend and which brands to trust.

The examples below are ours, written to show the shape of mattress questions; they are illustrative, not logged queries:

  • Sleeper fit: “Best mattress for a side sleeper who runs hot, queen, under $1,000.”
  • Type: “Hybrid or memory foam for a couple where one person moves a lot?”
  • Comparison: “Saatva vs Helix: which is better value?”
  • Trust: “Is this online mattress brand legit?” or “What do owners complain about?”
  • Terms: “Mattresses with the longest trial and free returns.”

Some questions touch health, such as back pain. Assistants may answer them, but a mattress brand should not try to win them with medical claims. Amerisleep itself argues that an AI model cannot see the variables that decide whether a mattress works, such as body weight, sleep position and temperature sensitivity. A reasonable expectation is that answers to health-tinged questions lean on medical and review publishers more than on brands; see what commercial health sites cited by ChatGPT have in common.

How does an AI answer turn into a mattress sale?

Through a short list that sends the shopper to a brand site or store, ready to buy.

The short list. An answer names a few mattresses. Given that a third of shoppers stall on choice, being named may be the difference between a purchase and another year on the old mattress.

The purchase. The shopper visits your site, a retailer or a showroom. Many mattress shoppers move between channels, so some AI-influenced sales will close in a store where no tracking links them to the answer.

The keep. If the answer matched the shopper’s needs, the mattress is more likely to stay. In the Amerisleep survey, 58% of buyers had at least one regret, and 18% said they chose the wrong firmness or comfort. We infer that clear, accurate firmness and fit information, both on your site and in what AI tools say about you, protects revenue after the sale as well as winning it.

The add-ons. Toppers, pillows and bedding follow a kept mattress. Brands that also sell bed frames or bedroom sets can borrow from how furniture brands win high-ticket AI orders.

What decides whether an AI assistant names your mattress?

Product data, reviews and reputation; the platform documents some of it, studies show the rest.

Documented by the platform. OpenAI says that when ChatGPT selects products (opens in a new tab), it considers structured metadata such as price and product description from first-party and third-party providers, plus other third-party content, and may weigh price, reviews and ease of use. It also says labels such as “Budget-friendly” are generated by ChatGPT and are not guarantees, and that reviews and ratings are not verified by OpenAI.

Observed in studies. An academic study of brand bias in AI recommendations (opens in a new tab) tested skincare, where buyers rely on reputation because they cannot judge quality before use, much as with a mattress. When products had the same specifications, well-known brands were recommended 100% of the time, but that dominance disappeared with less than a 0.1-star rating advantage for a competitor. The same study found that authority-style language, including fabricated clinical-evidence claims, broke the leader’s hold 50% to 73% of the time. That is a finding about how easily models can be swayed, not a tactic. Under the FTC’s health products guidance (opens in a new tab), claims about health benefits require competent and reliable scientific evidence. Our guide to skincare products in AI answers covers the same study and the same FTC rule from the skincare side.

Reputation questions lean on review sites. In our study of “is this brand legit?” answers, 88.0% of answers cited a review or complaint platform, and claims attached only to review platforms were negative 56.5% of the time. For online mattress brands, which shoppers often ask about by name, that makes complaint handling and owner reviews part of AI visibility.

Trust factors specific to mattresses. We infer that what helps an assistant name a mattress is what a careful shopper checks: a firmness rating on a stated scale, materials, height, trial length, return fees, warranty terms, delivery and old-mattress removal, and independent testing by reviewers who sleep on the product. For how ratings and prices compare with brand names in AI product picks, see what drives AI product recommendations.

What does it cost a mattress brand to be missing, or misrepresented?

Shoppers who never consider you, and shoppers who buy the wrong mattress because of a wrong fact.

We found no study that counts mattress sales lost to AI answers. The exposure is clear, though. With 37% of shoppers using AI tools in their research and a third stalling on choice, a brand absent from those answers is absent from a growing share of short lists. Direct-to-consumer brands built their reach on paid marketing; NPR reported that Casper’s founders spent hundreds of millions of dollars on it. Being named in a free answer is one way to depend less on that spend, though we cannot yet measure how much less.

Misrepresentation has its own cost. An assistant that states an outdated trial length, the wrong firmness or a complaint pattern drawn from old reviews can send away a buyer who would have kept the mattress, or attract one who will return it.

How does GEO work for a mattress brand?

By making your mattresses easy to understand, compare and verify, without promising placement or making health claims.

  1. Verifiable product facts. Firmness on a stated scale, materials and layers, height, weight limits, edge support, sizes, trial length, return fees, warranty and delivery terms, consistent across your site, product feeds and every retailer that sells you.
  2. Honest fit guidance. Pages that explain which sleepers each model suits and which it does not. Admitting a trade-off is more believable than claiming to suit everyone.
  3. Substantiated claims only. No claims about pain, health or sleep quality without competent and reliable scientific evidence. Unsupported claims are a legal risk and can surface in AI answers next to the evidence against them.
  4. Real reviews, gathered fairly. The FTC’s rule on fake reviews (opens in a new tab) bans fake reviews and incentives conditioned on a particular sentiment. Planted reviews can also distort AI answers; see how fake reviews affect AI recommendations.
  5. Independent testing coverage. Reviewers and publications that test mattresses in person and explain their methods.
  6. Reputation monitoring. Track what assistants say when shoppers ask whether your brand is legit, and fix the underlying problems they cite.

For where useful GEO ends and manipulation begins, see legitimate GEO versus manipulation.

What can’t the data yet say about selling mattresses through AI?

Which mattress brands AI assistants name today, how stable those picks are, and how many sales they drive.

The best mattress-specific data on AI use comes from a survey run by a mattress company. The Better Sleep Council figures predate the spread of AI assistants. Adobe’s conversion data covers all retail. The brand-bias study tested skincare in a controlled setting, not mattresses on live assistants. We found no independent, published measurement of which mattresses assistants recommend for common sleeper questions, and nothing that separates AI-influenced sales from returns. Any promise that a brand will become “the AI’s pick” for mattresses goes beyond the evidence.

Where should a mattress brand start?

With an audit of what AI assistants say about your mattresses, your trial terms and your reputation.

List the questions your buyers ask: sleeper type, budget, mattress type, brand comparisons and “is it legit.” Check which brands are named, which review sites are cited, and whether the facts about your firmness, trial and returns are right. Then fix the facts, the review gaps and the claims you cannot support. If you would rather have us run it, ask us for a mattress visibility review and we will show where your brand stands in AI answers for the questions that lead to mattress orders, and what would help more of those orders stay sold. For the ongoing work, our generative engine optimization service page covers keeping firmness, trial and review facts consistent wherever sleepers and assistants look.

Frequently asked questions

Can an AI assistant tell a shopper which mattress is best for back pain?

It may try, but a mattress brand should not rely on it or encourage it. Health claims need scientific evidence, and AI tools cannot assess a shopper’s body or condition. Shoppers with pain should ask a clinician.

Do mattress shoppers still start on Google?

Most do. In the Amerisleep survey, 49% started on Google, while 37% used AI tools at some point in their research.

Can we offer discounts in exchange for reviews?

Not discounts tied to a positive review. The FTC’s rule bans incentives conditioned on reviews expressing a particular sentiment, positive or negative.

Do longer sleep trials help with AI visibility?

We do not know of evidence that they do directly. Clear trial and return terms help shoppers and assistants compare offers, and fewer mismatched purchases mean fewer returns.

Sources

Free strategy call

Some questions are easier to answer about your own business.

Bring the one that matters most. On a free 30-minute call we’ll take a first look at it and send you a short written read afterward.

Ask about my businessHow we work