Industries · Home, garage & equipment
AI search visibility for home and garage retailers.
In this category the answer is a number. Will it run the fridge and the sump pump. Will the lift clear a ten-foot ceiling. Does it fit a 2015 F-150. Nobody asks an assistant for the nicest generator, and an assistant can only answer a constraint question if the numbers exist as text on a page it can actually read.
On most sites in this category they do not. The specs are a PDF, the fitment is a picker that resolves in the browser, and the sizing answer only exists inside a calculator. The page looks complete to a shopper and close to empty to the systems writing the recommendation.
Last updated 2026-08-22 · Deadpan Labs · Miami, FL
What buyers actually ask
Not one of these is a “best product” question.
Every one of them is a load, a clearance, a voltage, a fitment or a warranty term. That is what makes this vertical structurally different from the rest of ecommerce, and it is why the fix here is unusually concrete.
- “Will a 3,500-watt generator run my fridge, freezer, sump pump and router at once?”
- “What's the minimum ceiling height for a two-post lift — will one fit in a 10-foot garage?”
- “Does this lift need 220V, and what breaker does it want?”
- “Does this part fit a 2015 F-150 5.0L XLT?”
- “Is the five-year warranty residential-only? Does commercial use void it?”
A page carrying a price and a marketing paragraph is not a candidate for any of them. A page carrying the numbers, as text, usually is.
Your list is not this list. A panel is built from your loads, your fitments and your actual competitors, which is the first thing the free check produces.
The diagnosis
Six reasons an equipment retailer goes unrecommended.
The general reasons are on the pillar page. These six are the ones specific to spec-driven retail, and unlike most categories, five of the six are entirely within your control.
The crawlers do not run your JavaScript
A fitment picker is an empty page to the systems doing the reading.
Vercel's December 2024 analysis of AI crawler traffic found that the major AI crawlers fetch JavaScript files but do not execute them; Google's renderer was the exception. A “select your vehicle” widget, a specifications tab that loads on click, or a compatibility table fetched from an internal service therefore does not exist as far as most of these systems are concerned. The page looks complete to a shopper and close to empty to a parser, and this is the single most common defect we find in this category.
The specs are a picture of a table
Torque, CFM, wattage and dimensions shipped as a PDF or a JPEG.
A manufacturer spec sheet linked as a PDF, or a specification table exported as an image and dropped into the gallery, yields no extractable text at all. The retailer has the data, has published the data, and still cannot be cited for it. Whichever site did type the numbers out as text is the one that gets quoted — usually the manufacturer, a marketplace listing, or an affiliate buyer guide.
The answer requires arithmetic nobody published
“Will it run my fridge” is a sum, not a lookup.
Starting watts against running watts, CFM at ninety PSI against a tool's duty cycle, deck width against acreage, tonnage against square footage. These questions need two numbers and a calculation, and the calculation usually lives inside an interactive sizing tool that produces an answer only for someone who clicks it. There is no static page for an assistant to cite, so it cites a blog post instead — and the blog post recommends somebody else's product.
No identifiers, no entity
The engine cannot tell your listing is the product being reviewed.
Google documents name, image and offers as the required properties on a merchant listing, with GTIN, MPN, brand and SKU only recommended. Recommended is doing a lot of work in this category: without a stable identifier there is nothing tying your page to the reviews, the roundups and the spec sheets that discuss the same product elsewhere. The engine resolves the ambiguity by falling back to the manufacturer or the marketplace, which is a rational thing for it to do and a terrible outcome for you.
Variants that the schema does not model
Google supports six variant axes, and none of them is voltage.
Google's product variant documentation supports exactly six axes a product group may vary by: color, size, suggested age, suggested gender, material and pattern. This category varies on voltage, amperage, CFM, tank capacity, deck width, hose length and load rating — not one of which is on that list. So there is no clean way to model the variants this category actually has, and what most catalogs ship instead is twenty near-identical URLs with nothing connecting them at all. There is also no dedicated Google attribute for auto-parts fitment, despite what a great deal of agency writing claims — which means the most commonly sold answer to this problem is an attribute that does not exist.
Two retailers already own the category
The concentration inside the answers is tighter than it is in the market.
5W Public Relations' Home Services & Improvement AI Visibility Index, published in May 2026 against Q2 2026 data across sixty-plus homeowner prompts, put Home Depot at 16.5% and Lowe's at 13.5% of tracked citation share. An independent retailer is not going to take a general category question off either of them. What is winnable is the specific constraint question — the exact fitment, the exact load, the exact clearance — which the giants answer generically and you can answer precisely.
The part you control
Type the numbers out. Most of your competitors have not.
This is the rare vertical where the highest-leverage work is unglamorous and almost entirely inside your own catalog. It is also the rare vertical where the cause and effect is legible, because a spec either is retrievable as text or it is not.
Concretely: the numbers a buyer’s constraint question turns on — voltage, CFM, duty cycle, load rating, clearance, fitment — have to exist as retrievable text, tied to identifiers that connect your listing to everything written about the same product elsewhere. Most catalogs in this category fail on at least three of those, and which three changes what the work costs.
We check that against what a crawler can actually retrieve, not against what the page looks like when you load it yourself. The two are different far more often than owners expect, and our free crawler check will tell you for nothing whether the crawlers are even allowed to look.
None of that is a promise about a ranking or a recommendation. It is a statement about whether you are eligible to be considered at all, which is a lower bar and a necessary one.
Measurement
What a panel measures in this category.
A prompt panel is a fixed set of real buyer questions asked across the answer engines on a fixed schedule. Here it cannot sensibly be built from category head terms, because two national retailers already own those and an independent is not going to take one off them.
It has to be built from the questions that are realistically winnable instead, and working out which of yours those are is most of the difficulty. A panel full of terms you were never going to win reports a flat line every month and teaches you nothing about the questions you could have had.
We run these panels monthly, including on our own brands. That is not a case study in this category and we will not present it as one — we have no signed client work here to point at. It is how we know what the run-to-run variance looks like.
Who this is not for
Five reasons to not hire us.
- Your catalog is dropshipped with manufacturer-copied descriptions. If your page says exactly what four hundred other pages say, there is no version of this work that makes you the citable one.
- You cannot get product identifiers from your suppliers. Without them the entity problem above is not solvable from your side, and we would rather say that now.
- Your specs genuinely only exist as supplier PDFs and nobody on your side can extract them. That is a data project before it is a visibility project, and it should be priced as one.
- You want general category rankings for terms two national retailers already own. We will tell you that is not a realistic target rather than take the money.
- You want a guaranteed AI recommendation. Nobody can honestly promise that, in this category or any other.
Questions buyers ask us
Frequently asked.
Why do spec-driven categories fail differently from other ecommerce?
Because the buying question is a constraint problem rather than a preference problem. Nobody asks an assistant for the nicest generator; they ask whether a specific generator will carry a specific load. That question is answerable only if the relevant numbers exist as text on a page the engine can retrieve. In apparel or beauty an engine can produce a defensible answer from editorial and reviews alone. Here it cannot, so the sites that typed their numbers out win, and the sites that shipped them as a PDF do not.
Our fitment tool works perfectly. Why would that be a problem?
Because it probably works in the browser. Vercel's December 2024 analysis of AI crawler traffic found that the major AI crawlers fetch JavaScript files but do not execute them — Google's renderer was the exception. A picker that resolves fitment client-side, or fetches it from an internal service on selection, presents an empty shell to those systems. The tool can be excellent for shoppers and simultaneously invisible to the thing writing the recommendation, and the two facts are easy to confuse because the page looks fine when you load it yourself.
Is there a Google attribute for auto parts fitment?
No, and this is worth being precise about because a lot of agency writing says otherwise. Google's Merchant Center product data specification contains no dedicated fitment or compatibility attribute for parts; the vehicle attributes that do exist belong to vehicle ads for dealerships, which is a different product. The vehicle listing rich result has also been retired. There is no structured shortcut here, and anyone selling you one is selling you an attribute that does not exist.
How do we handle variants when the schema does not cover our axes?
Google's variant model supports six axes and none of them is voltage, CFM, capacity or load rating. There is no clean answer to that, only a correct one and several wrong ones — and the most common wrong one, twenty near-identical URLs with nothing connecting them, is what most catalogs in this category currently ship. Which of the options is right for you depends on how your catalog is actually organized, so it is a question an audit answers rather than one a page can.
Can an independent retailer realistically compete with the national chains here?
Not on the category question. 5W Public Relations' Home Services & Improvement index, published in May 2026 across sixty-plus homeowner prompts, put Home Depot at 16.5% and Lowe's at 13.5% of tracked citation share, and that concentration is tighter inside the answers than it is in the market. What is winnable is the constraint question — the specific fitment, the specific load, the specific clearance — because the national listings answer those generically and a specialist can answer them exactly. That is a narrower target and a much more defensible one.
What can Deadpan Labs honestly claim in this category?
That we measure it and run the work that changes it, and that the machine-readability half of this is the part with the clearest cause and effect of any vertical we work in. We have no signed clients here and no case study to show, and we are not going to invent one. Our own receipts are that we run our own products on the same platform we sell, that we took one of them from zero to more than 900 organic downloads with no ad spend, and that we run monthly AI-visibility prompt panels on ourselves.
Keep reading
The rest of it.
- /ai-search-visibility
AI search visibility for ecommerce
The pillar: what it is, how it differs from ranking, and the six general reasons a store goes unnamed.
- /tools/ai-crawler-check
Free AI crawler check
Whether the crawlers are even allowed to read your site, read straight from your robots.txt. No email required.
- /industries/supplements
Supplement & wellness brands
A different problem — there the constraint is on what the brand is able to say at all.
- /industries/beauty
Beauty & skincare brands
Where third parties write the answer and your shade data is unreadable.
- /pricing
What it costs
Published: the audit price, the monthly bands, and the factors that move them.
The free check
Find out if your specs are readable at all.
Give us your category and a few real constraint questions. We’ll run them against ChatGPT, Perplexity and Google AI and show you who gets named, and which pages the engines read to decide.