From Searching to Asking: Why Your Website Is Checked After the Shortlist

For more than thirty years I have watched buyers change the way they find a supplier. Catalogs gave way to websites. Websites gave way to search engines. Search engines gave way to phones. Every one of those shifts changed where the buyer looked. None of them changed what the buyer did once they got there: they looked through the options themselves, and they made up their own mind.

That part is changing now.

Buyers no longer only search. They ask.

And that one change moves the moment of decision to a place most companies are not watching.

Searching: the world we built for

Think about how searching works. A buyer types a few words into Google. They get a page of links. They open several tabs, compare, go back, try different words, compare again. Somewhere in that process they land on your website, and your website does the work. It explains what you make, shows why you are different and asks for the inquiry.

For twenty-five years, that is what we optimized for. Rank on the first page. Earn the click. Make the website convincing once the buyer arrives. The choice happened on your site, so your site was where the battle was won or lost.

The whole discipline of search marketing was built on one assumption: the buyer does the comparing.

Asking: the world we are now in

Now think about how asking works. A buyer opens ChatGPT, Gemini, Perplexity or Google’s AI Mode and asks a full question, the way they would ask a colleague. Not three keywords. A sentence, with their situation in it.

They get one answer. In that answer, two or three names.

The comparing has already been done for them. The AI read the sources, weighed the options and handed back a shortlist. The buyer did not open ten tabs. The buyer may not have opened any.

So when that buyer finally visits a website, they are not there to compare. They are there to confirm a choice that is mostly made.

Searching was a buyer making a list. Asking is a buyer receiving one.

How an AI answer is built

To see why this matters, it helps to look at how an AI answer comes together. When a buyer asks a question and the AI searches the web to answer it, the work happens in roughly three steps.

First, it reads the sources it trusts. For most buying questions, those are not company websites. They are review sites, publications that test products, comparison articles, “best of” lists, forums, videos and the other places where people who are not the seller describe what is good.

Second, it builds a shortlist. From those sources, it picks the two or three names that come up most often and most favorably for that specific question.

Third, it checks the websites of the brands on the shortlist. It looks at the product pages, the specifications, the prices, the policies, to confirm and fill in the details.

Read that order again. The websites are checked in step three. The shortlist is made in step two, from sources that are not yours.

The website is checked after the shortlist. First you have to be on the list.

This is the single most important idea in this article, so I will put it plainly. You can have the best website in your industry, fast, well written, technically perfect, and still be invisible in the answer. Not because your site is weak, but because the decision about who gets considered was made somewhere else.

Where the shortlist is decided

If the shortlist is not decided on your website, where is it decided?

It is decided by what others say about you. The review site that tested your product, or did not. The trade publication that included you in its roundup, or left you out. The comparison article that put you next to your competitors. The forum thread where a buyer asked for recommendations and someone answered. The YouTube video where a customer showed your product in use.

That last one surprises people. Searchable, an AI visibility platform, looked at 72.4 million AI answers from July to September 2026. About 11% of them used a YouTube video. For “how to” questions, Google’s AI answers used a YouTube video 45% of the time. And four in ten of the videos AI cited had fewer than 1,000 views. The video nobody watches can still be the source that puts a brand in front of a buyer.

None of this is new in principle. A good reputation has always come from what others say about you. What is new is that a machine now reads all of it, at once, every time a buyer asks a question, and turns it into a shortlist of two or three names.

That has a consequence most companies have not absorbed yet. Many of the sources AI reads cover what they cover. If an independent site reviewed your entry-level product but never tested your flagship, the AI can only recommend what it has read about. A missing review is not a neutral gap. It is a missing seat at the table.

Your website is still a witness

None of this means your website stops mattering. It changes jobs.

Your website is now one witness among many. The AI checks it in step three, and what it finds there either confirms the story the other sources told or confuses it.

Here is where many companies are exposed without knowing it. Every document that reaches a customer is a witness: the product page, the spec sheet, the instruction manual, the packaging, the FAQ, the listing on a retailer’s site. When those witnesses agree, the AI repeats your story with confidence. When they disagree, the AI decides which version your customers hear.

I see this more often than you would expect. A manual says one thing about a product. The website says another. A spec sheet says a third. Nobody inside the company noticed, because nobody reads all three side by side. The AI does. And it does not call your customer service line to ask which one is right. It picks one.

One brand. One truth. Every channel.

There is a second job your website still does well: it answers the questions buyers ask on the way to a decision. A page that answers a real buyer question clearly, in the words buyers use, gives the AI something to quote. A page that only describes the product in catalog language gives it very little.

What this does to your traffic

If AI now does the first part of the buyer’s journey, you would expect fewer people to visit your website. You would also expect the ones who do visit to be much further along. Both are showing up in the data.

In a 2025 study of real browsing data, Pew Research Center found that when Google showed an AI summary, people clicked a link on about 8% of visits, against about 15% when there was no summary. The answer is doing more of the work, so fewer people need to click.

Orbit Media looked at 97 Google Analytics accounts and found that AI tools sent only about half a percent of total visits. But those visitors converted to leads at roughly three times the rate of visitors from regular search.

Small traffic. High value.

That fits what I have believed from the start. AI is now doing the screening and the educating. The buyer who reaches your site has already been through the early questions, the comparisons and the shortlist. They arrive informed and with higher intent, and they convert.

So if your website traffic is flat or falling, do not panic yet. Fewer visitors is not the problem by itself. Look at what happens after the visit: the inquiries, the quote requests, the RFQs, the sales. And ask the harder question: what is AI telling buyers before they ever reach you?

One caution. A lot of this influence does not show up as “AI” in your analytics at all. A buyer reads a recommendation in ChatGPT, closes it, and a day later types your name into their browser. That visit shows up as direct traffic or as a search for your brand, not as a visit from AI. Orbit found that direct traffic now accounts for 34% of visits across 86 sites, ahead of organic search at 32%. The influence is real. It is just hard to see.

What the data says about AI and your traffic

Signal	What the data shows	Source
Clicks on a link when Google shows an AI summary	About 8% of visits, against about 15% with no summary	Pew Research Center, 2025, real browsing data
Share of total visits arriving from AI tools	About 0.5%	Orbit Media, 97 Google Analytics accounts
How those AI visitors convert	Roughly 3x the lead rate of visitors from regular search	Orbit Media, 97 Google Analytics accounts
Direct traffic against organic search	Direct 34%, organic search 32%	Orbit Media, 86 sites
AI answers built using a YouTube video	11% of all answers, and 45% of “how to” answers on Google	Searchable, 72.4 million AI answers, July to September 2026

Not every question matters equally

When companies first look at what AI says about them, most of them start by asking about themselves. They type their own brand name into ChatGPT and read what comes back. That is a useful check, but it measures the wrong moment.

Think about the order in which a buyer asks.

First comes the problem. “How do I stop this machine from shifting on an uneven floor?” No category yet, no brands. The buyer is learning.

Then comes the shortlist. “What are the best options for this job?” This is the moment that decides who gets considered. It is where you are discovered, or not.

Only then comes the brand question. “Is this company any good?” That question only exists after the shortlist has already put your name in front of the buyer.

Branded questions have always been the easy ones. In the search era, ranking for your own name was the easiest win in SEO. In the asking era, being named when someone asks about you by name is the easiest win too. But who is asking for your name if nobody knows it yet?

The questions that measure real progress are the ones without your name in them: the problem questions and, above all, the shortlist questions. That is where new buyers find you. It is also where your competitors are being recommended in your place.

So when you test, leave your brand name out. Ask the questions a buyer would ask before they have ever heard of you. Whether AI describes you correctly when someone asks about you directly is a separate test, worth doing, but it is not the one that brings you new business.

Where to start

The good news is that you do not need a new budget cycle to find out where you stand. You need a method, and a little discipline.

1. Listen to the questions your buyers actually ask. Not the keywords you wish they used. The questions your sales team hears on every first call. The emails and RFQs your customer service team answers every week. The words people type into the search box on your own website. Those are the questions buyers are now asking AI.

2. Ask them the way a buyer would. Use a private session with no history. Leave your brand name out, because a buyer who does not know you will not type it. Ask the same question three times, on more than one AI tool. One answer is an anecdote. Three start to show a pattern.

3. Write down who is on the list. Which brands are named, how often, and in what order. And just as important, which sources the answer was built from.

4. Earn your place where the shortlist is decided. If the AI reads three review sites and none of them has tested your product, that is your first project. If a trade publication’s roundup is cited in every answer and you are not in it, that is your second.

5. Make your witnesses agree. Take the three facts buyers care about most, such as the warranty, the dimensions or the standard your product meets. Check them in your spec sheets, your manuals, your product pages and your retailer listings. Fix every contradiction.

6. Do it again in thirty days. AI answers move. Competitors move. One run is a snapshot. The same questions, asked the same way every month, become a trend you can manage.

Built well for searching, not yet ready for asking

Let me be clear about one thing, because it gets lost in a lot of the noise about AI. None of this replaces good search work. AI answers lean heavily on pages that are already well built, well linked and well regarded. If your foundations are weak, they stay weak in the answer.

But strong foundations are no longer enough on their own. Most of the companies I talk to are built well for searching. Very few are ready for asking.

The difference is not a new technology. It is a new question. For twenty-five years we asked: how do we rank higher? The question now is: how do we get on the list?

---

Frequently Asked Questions: