We Audited 2,000 Websites in One Industry. Almost None of Them Could Close

This is a study about content that converts, and about how much of it the average company never actually makes.

Most content advice starts from the same assumption. Your problem is volume. You are not publishing enough, not publishing often enough, not feeding the machine fast enough.

I went looking for evidence that this was true, and found something quite different.

Earlier this year my team analysed close to 2,000 websites in a single industry, education, and scored each one on how well its content actually moves a stranger from mild interest to an enquiry.

We picked one vertical on purpose. Comparing a SaaS company to a law firm tells you very little, because the buyers, the budgets, and the timelines are completely different animals. Comparing 2,000 companies who are all selling roughly the same kind of thing to roughly the same kind of person tells you a great deal, because whatever separates them is a genuine difference in practice rather than a difference in industry.

Here is the top line. Adoption was nearly universal. Around three in five of those companies ran an active, deliberate content presence. Blogs, resource hubs, recognisable content types, a real publishing rhythm. By the usual measure of whether a company "does content marketing", the industry passed comfortably.

Then we sorted that content by the job it was doing, and the whole picture fell apart.

The content that converts is the content nobody makes

Across every active site we looked at, awareness content outnumbered decision-stage content by more than two to one.

That ratio sounds abstract, so here is what it looks like in practice. These companies had plenty of material for someone wondering whether they have a problem, and almost nothing for someone who has decided they do and is now choosing who to give money to.

The format-level numbers make it concrete:

• Pricing on fewer than one in four sites
• Case studies on about 35%
• Objection-handling content on about 30%
• Comparison pages on about 5%
• ROI calculators on about 3%

Every single item on that list is something a buyer goes hunting for in the last fortnight before they commit. And most of the market simply has not made them.

Take the pricing number, because it is the one people argue with most. The standard objection is that pricing is complicated, every deal is bespoke, and publishing a number would be misleading. Fine. But consider what happens to the buyer holding a shortlist of four suppliers. Three of them give some indication of cost. One makes them book a call to find out. That fourth company does not get dropped because it is expensive. It gets dropped because it is unknowable, and there are three others who are not.

Comparison pages are stranger still. Somebody, right now, is searching your brand name next to a competitor's. That page exists whether or not you write it. If you do not own it, the version your buyer reads was written by a competitor with every incentive to frame you badly, or by an affiliate who has never used either product. Roughly 95% of this industry has left that page to someone else.

The ROI calculator gap tells you something about internal politics that most marketers underrate. In any purchase above a certain size, the person who likes you is rarely the person who signs. Your champion has to walk into a room and justify the spend to someone in finance who has never heard of you. If you have not published anything that helps them build that case, you have handed your champion homework, and quite a lot of the time they simply do not do it.

Rolled into a single score, sales enablement came out at 1.57 out of 4. It was the weakest dimension in the entire study, by a clear margin. What that means in plain terms is that the average company here spends real money attracting attention, then hands that attention a website that cannot answer the questions people ask when they are ready to spend.

Most websites have nowhere to send anyone

It gets more basic than missing formats.

Conversion architecture, meaning the actual route from reading something to contacting a human, scored 1.88 out of 4.

About one in thirteen sites offered no visible next step from their content whatsoever. Someone reads to the bottom of a well-written article, feels vaguely persuaded, and finds nothing to do about it. That reader leaves, and the company never learns they existed.

Roughly 16% ran one generic call to action across the entire website. The same offer on every page, regardless of what the page was about or who was reading it. Only about 41% matched the offer to where the reader actually was in their thinking.

That last number is the expensive one, and it is worth sitting with.

If your only call to action is "book a demo" or "speak to sales", you are asking for the same level of commitment from everyone. From the person ready to sign this week, which is fine, and from the person who started looking four months ago and is still quietly educating themselves, which is not. The second group is much larger than the first. They are not ready to talk to a salesperson, and being asked to is often the moment they close the tab.

You lose them without ever knowing they were interested, which is why this particular leak is so hard to spot in analytics. It does not show up as a failure. It shows up as traffic that never converted, which most teams then try to fix by getting more traffic.

The AI problem almost nobody has staffed for

Now the part that genuinely surprised me.

AI Overviews now appear on roughly 85% of high-intent queries in this category, the kind of search someone runs when they are close to choosing. Traditional featured snippets show up on about 5%.

So for a growing share of buyers, the first impression of your business is not your website at all. It is a summary written by a machine, assembled from whatever it could find about you and whatever it decided to trust.

What those systems reward is demonstrable expertise. Named authors, visible credentials, original material, clear editorial ownership. Signals that a real person with real standing produced the thing. If you want to see how you currently surface, Lilach's AI visibility checker is a quick way to find out where you stand before you change anything.

Here is the number that made me sit up. Original research or proprietary data appeared on about 29% of the sites we looked at. Author bios carrying actual credentials appeared on about 26%. A stated editorial policy, roughly 3%.

Meanwhile, answer-engine optimisation showed up in about one in eighty of the marketing jobs this industry was hiring for. The same industry was naming TikTok in job adverts at roughly ten times that rate.

Read those two facts together. The thing AI search rewards most is close to the thing this industry produces least, and the skill required to fix it is the one almost nobody is recruiting for. That is not a crisis. It is an unusually wide door, and it is the same opening showing up across search more generally right now.

Why shared weaknesses are good news

Every gap in that study was widely shared. That is precisely what makes it useful.

When a weakness is universal, closing it does not mean out-executing a crowded field of people already doing the work well. It means competing against an absence. The comparison page nobody wrote. The proof nobody published. The pricing nobody would commit to.

The data supports that. Higher-growth companies in the sample scored better on every dimension we measured, and the margins were consistent rather than spectacular. Active content presence at 52% against 43% for the low-growth group. Conversion architecture at 1.94 against 1.85. Sales enablement at 1.66 against 1.52. Clarity about who they were for at 2.55 against 2.37.

None of those gaps is dramatic on its own. What matters is that all four move together, which suggests the companies pulling ahead are not winning on one clever tactic. They have closed the whole loop, and the loop compounds.

The other finding worth knowing is what did not predict maturity. Funding status barely moved the numbers at all, while revenue band roughly doubled active-presence rates between the smallest companies and the largest. Capital does not buy this. Operating discipline does, which is genuinely good news if you are the smaller party in your market.

And the raw material is usually already sitting in the building. The testimonials are in somebody's inbox. The results are in a spreadsheet. The objections live in the head of whoever answers the phone, who could recite the top three from memory. Almost none of it ever reaches the website, which is the real reason so much content stops working the moment it is published.

The full education study is here, and we ran a cross-industry version asking the same questions across B2B if you want to check whether your own sector looks any healthier. It mostly does not, which I found oddly comforting.

Stefan Kalpachev is the founder of Content RevOps, where he builds content and demand systems for B2B companies in complex, trust-led markets. He publishes first-party research on how industries market and convert, including The State of Content Marketing for Education Companies in 2026.