Anonymised, Not Empty: A Monthly Ritual for the Search Queries You'll Never See

Every few weeks someone posts the same screenshot in a UK ecommerce Slack group: a Search Console query table with a fat grey band labelled "(other)" sitting at 40% of clicks, and a message underneath saying "is this normal or have I broken something?"

You haven't broken anything. The bucket is doing exactly what it was built to do, and it will keep growing as your site picks up more long-tail traffic. The trick is to stop treating an anonymised report as a defective one and start reading the pattern underneath it.

Why the row exists at all

Google holds back a query when too few people typed it during the reporting window, or when typing it out could plausibly point to an individual — think health questions, address-style searches, anything with a name in it. It's a privacy threshold rather than a fault, and it applies across the whole site.

Two things follow from that, and both matter for UK shops.

First, the totals still reconcile. Those clicks and impressions are counted, included, and baked into every site-level number you look at. What you lose is the label, not the visit.

Second, the size of the bucket scales with your tail. A national retailer bidding on "car insurance" will see a tiny (other) share because its traffic is dominated by a few enormous head terms. A Leeds independent selling 400 different ceramics will see a much bigger one, because its traffic is spread across hundreds of tiny, specific searches. A high share isn't a warning sign — it's often a description of your business model.

The share also swells at predictable moments. Late November and early December, half-term weeks, bank holiday weekends: that's when people search in one-off, oddly phrased ways. If (other) jumps in the run-up to Christmas, that's demand doing something seasonal, not analytics doing something wrong.

What's hidden and what isn't

It's worth being precise, because the vague anxiety around this row is worse than the row itself.

Hidden from you: the wording, the exact query-to-page pairing, the individual position for each of those searches.

Still visible: how many clicks and impressions those searches generated, which landing page caught them, which country and device they came from, and how the numbers move month to month.

That's a lot to work with. You're simply moving from keyword-level analysis to page-level analysis — which, quietly, is the more durable way to look at demand anyway. Keywords rotate with the seasons. Intent sits still.

Step one: measure the shape, not the words

Once a month, export the Pages table from Search Console and add two working columns of your own: total clicks, and what proportion of those clicks landed in (other). Do it in a spreadsheet, not in your head.

Sort by that share. The pages sitting well above your site median are your long-tail entry points — the ones being found through dozens of small queries rather than one big one. Then compare against last month. A category page drifting from 30% to 55% is telling you something has changed in how it's discovered, even if it won't say what.

Step two: read landing pages like sentences

Take your flagged pages and look at three lines together: impressions, average position, and (other) share. If impressions climb while the share climbs, you're being found by lots of small, specific searches — a healthy signal for a specialist retailer, and a strong hint about which phrases belong in your page titles and H2s.

If impressions are flat but the share rises, your visible queries have shrunk. That's usually a sign the page is being discovered on thinner, more scattered terms than before, which may be worth a content refresh.

Step three: borrow eyes from elsewhere

You have two useful cross-checks that don't cost anything extra.

Bing Webmaster Tools will often show you queries Google withholds. Bing's slice of UK search is modest, but the phrasing of long-tail queries overlaps heavily with Google's, so it acts as a decent translation layer. Meanwhile, if you're running Google Ads, the search terms report (however truncated it has become) tells you which wordings are converting for you commercially.

Neither is a substitute. Both are useful triangulation.

Step four: logs and your own search box

Server logs won't recover the query — Google strips it before the referrer ever reaches you — but they'll confirm which landing pages are absorbing those dispersed arrivals, and at what times of day. Pair that with your on-site search terms. What people type into your own search bar is often a close cousin of what they'd type into Google, and unlike the (other) bucket, you can read every word of it.

The twenty-minute monthly loop

• Export the page-level table, calculate (other) share per page, sort.
• Flag everything above your own median — don't borrow someone else's threshold.
• Pick three movers and note them in a shared doc.
• Check impressions and position on those three pages.
• Cross-reference the intent cluster in Bing and in paid search terms.
• Write one plain-English line per page — "found via lots of small searches about X" — and pass it to whoever writes your titles and meta descriptions.

That last step is the one people skip, and it's the one that turns a report into a decision.

A quieter way to think about it

The (other) row will never be filled in. Chasing it is a losing game, and buying a third-party tool that claims to unmask it is usually a way to spend money confirming what your landing pages already told you.

Ask a different question instead: which doors are people walking through, and are those doors in the right place? Answer that each month and you'll know more about your demand than any keyword list would have told you — and you'll stop flinching at the grey band.