.svg)
When we put 10,580 buyer questions to 529 B2B websites, I expected the headline to be silence: questions the site never answers, or answers it to "contact sales". That is not what came back. Websites said nothing at all to 6.3% of the questions and sent 1% to sales. The big number was in between. One question in three got half an answer.
I think the half-answer is the most expensive gap on a B2B website, because nobody can see it. A missing page shows up in an audit. A "contact us" shows up in a funnel report. A half-answer looks like an answer to the person who wrote the page, and like a reason to open the next tab to the person reading it. This report is the full set of results from the Storylane Buyability Index, the method behind them, and where the half-answers cluster.
How the Storylane Buyability Index works
A scan takes one company's website and does what a careful buyer would do before booking a call. It writes 20 questions a buyer would plausibly ask that company, based on what the company sells and who it sells to. Then it tries to answer each question from the company's own site, and keeps an answer only if it can quote the supporting text from the site.
Each question gets one of five outcomes:
- Fully answered: a concrete answer backed by a quote found word for word on the company's site, with its claims checked against the page it came from.
- Partly answered: the site had something, but it was incomplete or failed a verification step.
- Gated, fairly: an enterprise topic, such as a custom SLA or a data processing agreement, where "talk to us" is normal.
- Sent to sales: "contact us" where a straight answer was expected.
- Not on the site: nothing found.
Three caveats apply to every number below, and to every post that quotes this data:
- The questions are generated per company. They are what a buyer would plausibly ask, not questions logged from real buyers.
- The 529 sites were chosen by the people who ran scans. This is not a random sample of B2B websites.
- Scans ran from Sep 8 to Oct 20, 2026. Each site counts once, using its most recent complete scan. Low-confidence scans, sites that blocked the crawler and sites that scored zero were left out.
Topic and page-type groupings below use keywords and URL patterns. They are good for the shape of the data, not for exact counts. You can run the same scan on your own site with the Storylane Buyability Index.
How often do B2B websites answer buyer questions?
A little under six times in ten. Across 10,580 questions on 529 B2B websites, the site gave a full answer 58.6% of the time and a partial answer 33.1% of the time.
| Outcome (Storylane Buyability Index, 529 sites) | Questions | Share |
|---|---|---|
| Fully answered | 6,204 | 58.6% |
| Partly answered | 3,506 | 33.1% |
| Gated, fairly | 106 | 1.0% |
| Sent to sales | 102 | 1.0% |
| Not on the site | 662 | 6.3% |
Two things stand out. First, B2B websites rarely refuse. Gated and sent-to-sales answers together are 2% of all questions. Second, the site said something to 91.8% of questions. The problem is not that sites are empty. It is that what they say often stops short of the answer.
The spread between sites is wide. Every site got 20 questions. The median site fully answered 12. The bottom 57 sites (about one in nine) fully answered six or fewer, and 88 sites fully answered 16 or more. Only 2 of 529 sites fully answered all 20. And 242 sites, 45.7%, had at least one question where the buyer would find nothing at all (we looked at those dead ends on B2B websites in detail).
Why are half-answers the real gap?
The obvious explanation for an unanswered buyer question is that the company is hiding something. Pricing is the usual example. That explanation fits a small slice of the data. Pricing does account for 90 of the 102 sent-to-sales answers. But sent-to-sales is 1% of everything. If companies were withholding answers on purpose, we would see far more gating and far fewer half-answers.
I have a theory for why the half-answer dominates instead. A half-answer is usually the general version of a specific fact. To take illustrative examples: "Fast setup" instead of "four to six weeks for a team of 50". "Enterprise-grade uptime" instead of the SLA number and what happens when it is missed. "Flexible plans" instead of the minimum contract. The general version is easy to approve, because it commits nobody to anything. The specific version needs someone in product, legal or finance to sign off on a number. So the page ships with the adjective, and the number stays in a sales deck.
That is why I call it the half-answer gap: B2B websites rarely refuse a buyer's question; they answer the safe half of it and leave out the specific half the buyer came for. The specific half is what a buyer needs to defend a shortlist to their boss, and it is the part the buyer cannot get without a call.
One honest limit on this theory. "Partly answered" also includes answers that failed a verification step, so some half-answers are complete answers the scan could not fully confirm. I think the pattern still holds, because the half-answers are not spread evenly. They pile up on exactly the topics where a specific answer means a commitment, as the next section shows.
Which buyer questions do websites answer worst?
By category, product questions do best and support questions do worst. The order is close to a ranking of how much commitment a full answer requires.
| Category | Questions | Fully answered | Partly answered | Gated or sent to sales | Not on the site |
|---|---|---|---|---|---|
| Product | 1,945 | 76.7% | 20.8% | 0.1% | 2.5% |
| Integrations | 1,019 | 71.4% | 23.2% | 0.0% | 5.4% |
| Comparison | 913 | 69.9% | 22.3% | 0.2% | 7.6% |
| Security | 1,587 | 58.6% | 33.7% | 1.3% | 6.4% |
| Implementation | 1,252 | 49.8% | 42.4% | 0.5% | 7.3% |
| ROI | 1,033 | 47.1% | 46.3% | 0.3% | 6.3% |
| Pricing | 1,530 | 46.4% | 35.2% | 10.3% | 8.1% |
| Support | 1,127 | 43.7% | 48.2% | 1.1% | 7.0% |
A further 174 questions did not fit a category and are left out of this table. Look at the partial column. Support, ROI and implementation questions are partly answered 42% to 48% of the time. Those are the topics where the honest answer is a number with conditions attached.
Narrower topics, grouped by keywords in the question, make the pattern sharper. These are the ten worst-answered topics asked on at least 50 sites:
| Topic | Questions (sites) | Fully answered | Partly answered | Not on the site |
|---|---|---|---|---|
| Support SLA and uptime | 235 (232) | 25.1% | 65.5% | 6.4% |
| Support hours and response time | 128 (122) | 38.3% | 55.5% | 5.5% |
| Contract and cancellation terms | 712 (430) | 39.6% | 39.6% | 12.4% |
| Per-seat pricing | 130 (116) | 40.8% | 36.2% | 10.0% |
| ROI and payback | 164 (163) | 40.9% | 54.3% | 4.9% |
| Implementation time | 957 (502) | 43.2% | 49.0% | 7.5% |
| Where data is stored | 360 (356) | 44.4% | 51.7% | 1.9% |
| Discounts | 90 (89) | 48.9% | 36.7% | 8.9% |
| Annual vs monthly billing | 196 (170) | 50.5% | 35.7% | 6.6% |
| ISO 27001 | 58 (58) | 56.9% | 29.3% | 12.1% |
Only one SLA question in four got a full answer, and two in three got half of one. The best-answered topics sit at the other end: API questions were fully answered 75.8% of the time, free trial questions 73.7% and SSO questions 68.6%. Those are yes-or-no facts that cost nothing to publish.
A note on implementation time, because two of our posts quote it differently. The topic above (957 questions on 502 sites, 43.2% fully answered) counts every question that mentions implementation, onboarding or setup. The narrower set that asks how long setup takes (529 questions on the same 502 sites) got a full answer 39.9% of the time. That narrower figure, rounded to 40%, is the one in our piece on why buyers ask how long before how much. Both are right; they measure different sets of questions.
Which buyers get the fewest full answers?
Each question is tagged with the buyer roles likely to ask it, and most questions carry more than one tag. End users and technical evaluators do best. Procurement and finance buyers do worst.
| Buyer role | Questions | Fully answered | Partly answered | Gated or sent to sales | Not on the site |
|---|---|---|---|---|---|
| End user | 5,068 | 65.4% | 29.5% | 0.2% | 4.9% |
| Technical evaluator | 3,727 | 63.8% | 30.9% | 0.5% | 4.8% |
| Security and compliance | 2,170 | 60.0% | 32.9% | 1.2% | 5.9% |
| Executive | 4,637 | 52.4% | 37.1% | 3.3% | 7.2% |
| Procurement and finance | 3,973 | 48.0% | 38.6% | 4.8% | 8.6% |
The scan also tags each question by how much assurance the buyer needs. Self-serve questions (does it do this, can I try it) were fully answered 69.0% of the time. High-assurance questions (the ones someone has to defend in a contract or a security review) were fully answered 49.8% of the time. The gap holds for every role, by 14 to 22 points: executives, for example, got 63.5% on self-serve questions and 44.5% on high-assurance ones. We wrote this up as the assurance gap: the more a buyer needs to be sure, the less the website says. (234 questions carried neither tag and are left out of that split.)
This matters more than it looks, because the people with the high-assurance questions are often not on your sales calls. In internal Storylane deal data across 637 B2B deals, four in five members of the buying committee never joined a call. The website is where they look.
Where on the website do the answers live?
Mostly far from where buyers start. Of the 6,204 full answers, the homepage supplied 113 (1.8%) and the pricing page 369 (5.9%). The blog, help centre and docs together supplied 2,352 (37.9%).
| Page type | Full answers | Share |
|---|---|---|
| Blog and resources | 872 | 14.1% |
| Help centre and FAQ | 850 | 13.7% |
| Docs and developer pages | 630 | 10.2% |
| Trust, security and legal | 622 | 10.0% |
| Product and feature pages | 478 | 7.7% |
| Pricing page | 369 | 5.9% |
| Case studies | 353 | 5.7% |
| Integrations pages | 223 | 3.6% |
| Comparison pages | 159 | 2.6% |
| Homepage | 113 | 1.8% |
| Other pages (not classified) | 1,535 | 24.7% |
A quarter of full answers came from pages the URL patterns could not classify, so treat the smaller shares as rough. The direction is still clear, and we went through it page type by page type in why most answers live in the docs, not on the homepage. This is answer distance: the answers often exist, but several clicks from the page the buyer landed on. A scanning agent with no time limit finds them. A buyer with three tabs open may not.
How do buyers phrase their questions?
Like people trying to picture life after signing. The median question is 11 words long. 32.1% start with "how", and 32.6% use "we", "our" or "us": our data, our team, our contract.
The clearest signal is in the first two words. 550 questions start with "how long" and 128 start with "how much". That gap is partly phrasing, since cost comes up in plenty of questions that start differently, but it says something about what buyers worry about once a product looks relevant. They want to know what it will take, not only what it will cost. Questions that start with "how long" got a full answer 39.3% of the time.
Pricing still matters, and it is where the "talk to us" habit lives: pricing is 14.5% of the questions but 158 of the 208 gated or sent-to-sales answers. We looked at that trade-off in whether you should put pricing on your website.
What should you do with this?
Start with your own 20 questions. Take the questions your sales team hears most, or run a scan, and grade each answer on your site the way the index does: full, partial or nothing. Then fix the partials before you write a single new page. A new page adds to the pile. Turning "fast setup" into "most teams go live in X weeks, here is what takes longest" turns a half-answer into a full one where the buyer is already reading.
For each partial, ask three things:
- What is the specific half? Usually a number, a condition or a timeframe.
- Who has to approve it? If the answer is "legal" or "finance", that is the real blocker, and it is a meeting to book, not a page to write.
- Where would the buyer look first? Put the answer there, not three clicks away in the docs.
Then think about the buyer who never reaches the page. More of them now arrive from an AI answer with a shortlist already open, a pattern we covered in whether your website can answer a buyer who arrived from ChatGPT. Those buyers ask in sentences, about their own situation, and they leave when the answer is vague.
This is where I should say plainly that I have a stake in it. We build RepX, a website agent that answers buyer questions on the spot, shows the product, works out whether the buyer is a fit and hands sales-ready buyers to your team. It learns from your website, sales transcripts, FAQs, files, demos, slides and videos, which is where this data says the answers already live, and qualified leads flow to HubSpot, Salesforce and Slack. It is best for teams whose answers exist somewhere in their content but sit far from the buyer. If you want to see how it would handle your 20 questions, book a demo.
The one idea to keep: your website probably does not refuse buyers. It half-answers them. Find the specific half, get it approved and put it where the buyer is already looking.