Website traffic down by a third, and the phone still ringing at the same rate. That is the message we get most often from business owners right now, usually with a screenshot attached and one question underneath: should I be worried?
It is a fair question. For fifteen years the industry told you traffic was the scoreboard. More visitors, more enquiries, more revenue. For a long time that was roughly true.
That link has weakened. Not broken, weakened. And owners who panic about it tend to make an expensive decision: cutting the marketing that still works, or paying someone new to chase a number that no longer pays anybody’s wages.
Here is what actually changed, how to tell whether your own drop matters, and what to do about it.
TL;DR: Website traffic down while enquiries hold steady is almost always a change in who is clicking, not a fall in demand. Google now answers the simple questions on its own results page, so research visits disappear first, and those visits were never going to buy. Judge the business on enquiry volume, enquiry quality and cost per enquiry over a rolling twelve months. If those are flat or improving, you have lost visitors, not customers.
Website traffic down does not mean demand is down
Every website mixes two audiences into one number: people who might buy from you, and people looking something up.
The second group has always been the larger one. They wanted a definition, a size, a rough price. They read a paragraph and left. They inflated your traffic and never appeared in your sales figures.
That group is now served before they reach you. So when a business tells us sessions fell from 4,000 a month to 2,600 while enquiries held at 45, we do not see a problem. We see the same 45 buyers with 1,400 fewer bystanders around them, and a cost per enquiry down by a third. The mistake is treating both audiences as one asset. They never were.
What changed: the answer now sits above the link
Google now writes a short answer at the top of the results page, above the list of websites. If it is good enough, nobody scrolls.
The clearest independent measurement comes from the Pew Research Center, which tracked the browsing behaviour of 900 adults across 68,879 Google searches in March 2025. Where an AI summary appeared, 8 percent of visits led to a click on a normal search result. Where none appeared, 15 percent did. Roughly half the clicks, gone.
Two further figures matter more. Only 1 percent of visits involved clicking a link inside the summary, so being cited is not the same as being visited. And 26 percent of people ended their session after a page with a summary, against 16 percent without. Pew has published the full breakdown. Google disputes it, calling the query set unrepresentative. But the direction of travel is not seriously contested, including by Google.
The studies contradict each other, so ignore the headlines
Here is the part nobody selling you services mentions. The research disagrees with itself.
Search Engine Land put the major studies side by side and found them pulling in opposite directions. Ahrefs measured roughly a 34 percent loss of clicks where an AI answer appears. Semrush recorded a slight decrease in searches ending without a click. On visitor quality, Amsive and Ahrefs found AI traffic converts better than Google, an ecommerce analysis found it converts worse, and a well known conversion specialist called it lazy, unqualified traffic.
The scale claims deserve the same scepticism. Semrush projects that the average AI search visitor is worth 4.4 times an ordinary organic one, and admits this is an extrapolation. WebFX examined 2.3 billion sessions to December 2025 and found AI referrals had grown 796 percent while still making up 0.18 percent of the total.
Put simply, the effect on you is not the traffic AI sends. It is the traffic it stops sending. No industry average answers your question. Only your own numbers can.
The four numbers that actually settle it
Put the traffic graph away. Pull these four instead, on a rolling twelve months.
- Enquiry volume. Forms, calls and emails that reached a human, counted the same way every month.
- Enquiry quality. What share became a real opportunity worth quoting for? A yes or no on each enquiry for a month tells you plenty.
- Cost per enquiry. Total marketing spend divided by enquiry volume. This says whether losing visitors made you poorer or richer.
- Average order value from web enquiries. Traffic can fall while the work gets bigger. That is a good year dressed up as a bad chart.
Most owners can assemble all four from their CRM and bank statements in half a day. With website traffic down and these four steady, you have your answer. We also set out what a website should be worth to the business.
When website traffic down really is a problem
Sometimes the drop is exactly what it looks like. The warning signs are specific.
- Enquiries have fallen in step with traffic across two or more quarters, not one soft month.
- The pages losing visitors are your service, product and pricing pages, not your blog. Nobody asks an assistant to summarise a quote form.
- You have lost visibility on the searches that describe what you actually sell, not the general questions around it.
- Branded searches are falling too. If fewer people type your company name, that is a demand problem and no technical work fixes it.
- A competitor is named in the AI answers for your core services and you are not.
Any two of those together justify spending money. None of them, and your problem is a chart, not a business.
What Google itself tells site owners to do
There is a small industry selling special preparation for AI search, so it is worth knowing what the people running the system say. Google’s guidance for site owners states plainly that there are no additional requirements to appear in its AI answers and no special optimisations necessary. No machine readable files, no AI text files, no extra markup. The advice is what it has been for a decade: meet the technical basics, publish useful content written for people, and let the site be crawled.
Google also claims clicks from pages carrying an AI answer are higher quality, in that those visitors stay longer. Treat that as an interested party’s claim. If someone quotes you for an AI optimisation package on top of your existing work, ask what it does that good content and a sound website do not. Make them answer in enquiries.
What to do in the next 30 days
- Get the four numbers. Volume, quality, cost per enquiry and average order value, twelve months rolling. Everything else waits.
- Split your pages into two lists. Pages that earn money, and pages that answer questions. Check which list your lost visitors came from. If it is the second, relax.
- Ask the assistants about your market. Type the five questions your best customers ask into ChatGPT, Gemini and Google, and see who gets named. Ten minutes, no software. Our guide to measuring your AI search visibility covers doing it consistently.
- Fix the commercial pages first. Clear pricing guidance, real proof, a fast enquiry route. Fewer visitors makes converting them worth more, not less.
- Only then reconsider the budget. If cost per enquiry improved, spending less is an odd response. What a business your size should actually spend beats a percentage quoted at a networking event.
The businesses that do well over the next two years will not be the ones chasing traffic back. They will be the ones who worked out which visits were ever worth having.
Is website traffic down a sign my SEO has stopped working?
Not on its own. Website traffic down while enquiries hold steady usually means you have lost research visits rather than buyers, because Google now answers simple questions itself. SEO has genuinely stopped working when you lose visibility on searches describing what you sell, when enquiries fall alongside traffic for two or more quarters, or when competitors are named in AI answers for your core services and you are not.
How much of a traffic drop is normal in 2026?
There is no reliable industry figure, and anyone quoting one confidently is guessing. The studies disagree sharply: Ahrefs measured around a 34 percent loss of clicks where AI answers appear, while Semrush found a slight decrease in searches ending without a click. Sector variation is wider than the average itself. Judge your own website traffic down on whether enquiry volume and cost per enquiry moved with it.
Should I still invest in SEO if AI answers the questions?
Yes, but judged on enquiries rather than rankings or sessions. AI assistants build answers from the same pages Google ranks, so work that earns visibility in one tends to earn it in the other, and Google’s guidance says no separate optimisation is required. What changes is the measurement. Pay for work that moves enquiry volume and cost per enquiry, not for reports about positions and impressions.
How do I tell if my enquiries are getting worse, not just fewer?
Score every enquiry for one month as a real opportunity or not, using whatever test your sales team applies. Compare that share against the same month last year, then compare average order value from web enquiries. Fewer enquiries at a higher qualification rate and a higher value is a stronger business. Falling volume with a falling qualification rate is the combination worth spending money on.
If you want a straight answer on whether your marketing is actually making you money, that is the conversation we have with business owners every week. Get in touch with Damteq and we will go through your numbers with you.
