Bing Sends Half Our Search Traffic. We Missed It.

We spent a day auditing twelve months of search data across 33 domains we own or manage. Then we found the audit had measured one search engine.

Search Console reports Google. Only Google. We knew that in the abstract and still wrote a report called “the estate”, built the headline table on it, and concluded that one of our directories was close to dead.

That directory was getting 96% of its search traffic from an engine we had not looked at.

This is what the other engines were doing.

The number that should not have been a surprise

Across 17 properties with measurable traffic, 90 days, by the engine that sent the session:

Bing delivered 49% of Google’s session volume. Non-Google engines were 46% of all search sessions.

The usual quoted figure for Bing’s UK market share is somewhere between 3% and 5%. Our portfolio skews towards local service sites and directories, which is precisely the profile where Bing over-indexes, so we are not claiming 49% is universal. We are saying that if you have never checked, your number is not the market share figure either.

DuckDuckGo converts at nearly twice Google’s rate

16.8% against 9.1%. A factor of 1.84, on a tenth of the volume.

That holds when you strip out the dross. Measured against engaged sessions only, so sessions where someone actually did something, DuckDuckGo runs 25.2% against Google’s 17.1%.

179 sessions is a small sample and we’ll say so before anyone else does. But 30 key events from 179 sessions is not a rounding error, and the direction has been consistent across three separate pulls.

Why it might be real: DuckDuckGo users have changed a default. Changing a default is a deliberate act, and people who make deliberate choices about their tools tend to arrive with more intent than people who typed into whatever box was in front of them.

AI assistant traffic converts as well as Google does

This one we did not expect.

221 sessions from ChatGPT, Perplexity, Copilot and Claude in 90 days, converting at 17.0% of engaged sessions against Google’s 17.1%.

Effectively identical. The volume is a tenth of Google’s and climbing. Every argument we have made about AI search being a real acquisition channel has rested on citation share and impressions. This is the first time we have been able to say the traffic converts at the same rate as the channel everyone already budgets for.

Small sample, again. We’re publishing it anyway because it’s the most interesting number in the set, and we’d rather be early and caveated than late and certain.

The worst case in the portfolio

One healthcare directory, over 90 days:

• Bing: 575 sessions
• Google: 34 sessions
• Bing produced 44 of its key events

Seventeen times more traffic from Bing than Google, and not one query, impression or position for any of it, because the site is not verified in Bing Webmaster Tools.

Our original audit looked at that property, saw 16 Google clicks in 90 days, and filed it as a problem site. We were judging it on 4% of its search traffic.

What we got wrong, specifically

Worth being precise, because the mistake is the useful part.

We used Search Console as a proxy for search. It is a proxy for Google, and everything downstream inherited that.

A click-through-rate table comparing site types understated every property whose audience is not on Google. A conclusion that directories convert badly was reached with their best engine missing from the data. And a “striking distance” list of page-one queries earning no clicks was fine as far as it went, but it was one engine’s page one.

We had not verified a single site in Bing Webmaster Tools. Not the healthcare directory on 96% Bing, not the dental directory on 58% non-Google, not our own site. The tool is free and has an Import from Google Search Console option that verifies and backfills in minutes.

We reported sessions when we should have reported engaged sessions. Roughly a quarter of the portfolio’s sessions came from cloud datacentre regions, and on two properties it approached half. Any conversion rate calculated on all sessions is wrong by that much, which is why both columns are in the table above.

What to do about it, in about ten minutes

Check your own engine mix. In GA4, make a report with Session source as the dimension and Sessions plus Key events as the metrics, over 90 days. Read two things: how much of your search traffic is not Google, and the conversion rate per engine rather than the volume. Most people have never looked.

Verify in Bing Webmaster Tools. Free, and the Search Console import does it in a couple of minutes. It covers Yahoo as well, and partly feeds DuckDuckGo.

Switch your reporting to engaged sessions. If Direct traffic shows high volume with low engagement and a single-digit average time on page, you are reporting bots to yourself.

Then stop calling a Search Console report an SEO report. Call it a Google report, because that is what it is, and go and find the rest.

Why we publish this sort of thing

Because the alternative is a case study with the failures edited out.

We have published 6,482 page-one rankings that produced 218 clicks and an experiment we ruined by treating all ten pages at once. This one belongs in the same series: a real dataset, a quotable number, and something that did not work.

If you want the method rather than our numbers, our page on the AI SEO framework sets out what the research actually proves and what it does not.

Common questions

Does Google Search Console show Bing data?

No. Search Console reports Google only, and there is no setting that changes that. Bing, Yahoo, DuckDuckGo and Ecosia are invisible to it.

If your SEO reporting comes from Search Console, it is a Google report regardless of what the title says. Bing has its own free equivalent, Bing Webmaster Tools, which also covers Yahoo and partly feeds DuckDuckGo.

How much traffic does Bing actually send?

Across our own portfolio of 17 measurable properties over 90 days, Bing delivered 1,084 search sessions against Google’s 2,190, so 49% of Google’s volume. Non-Google engines accounted for 46% of all search sessions.

That’s our portfolio rather than a universal figure, and it skews towards UK local and directory sites, but it’s a long way from the 3% to 5% market share figures usually quoted.

Which search engine converts best?

In our data, DuckDuckGo, by a wide margin. It converted at 16.8% against Google’s 9.1%, which is 1.84 times the rate, on a tenth of the volume.

Measured against engaged sessions rather than all sessions the gap holds: 25.2% against 17.1%. Bing converted at 6.1%, Ecosia 7.4% and Yahoo 4.7%.

Does traffic from ChatGPT and other AI assistants convert?

Yes, and better than we expected. Across the portfolio, AI assistant referrals produced 221 sessions in 90 days converting at 7.7%, and against engaged sessions they convert at 17.0%, effectively identical to Google’s 17.1%.

The volume is small and growing. The quality is not the problem.

How do I check my own engine mix?

In GA4, build a report with Session source as the dimension and Sessions plus Key events as the metrics, over 90 days. Look at how much of your search traffic is not Google, and at the conversion rate per engine rather than the volume. It takes about five minutes.

Why does Bing matter more than its market share suggests?

Two reasons. It is the default on Windows and Edge, so it collects users who never changed a setting, which skews older and more local in the UK.

And it supplies results to other surfaces, including Yahoo and in part DuckDuckGo, so one engine’s index reaches several front ends. A low headline market share does not mean low traffic for your particular site.

Is this finding specific to your sites?

Partly, and we wouldn’t claim otherwise. Our portfolio skews towards UK local service sites and directories, which is exactly the profile where Bing over-indexes. A global B2B SaaS audience would likely look more Google-dominated.

The transferable part is the method rather than our percentages. Check your own mix.