Clicks dropped. Now what? A 15-minute investigation from your phone

The notification says a property’s clicks are down a third. You have fifteen minutes before the next thing, a phone, and a client who may or may not have noticed yet. This is the order to work in.

Throughout, the numbers are from QueryWatch’s Demo Mode dataset for a sample property called brightloom.co, so you can follow the same screens in the app without connecting an account. They are demo numbers, not a real customer’s.

Minutes 0 to 2: is it real?

Two things make a drop look bigger than it is, and both are worth two minutes before you do anything else.

First, check which day you are looking at. Search Console performance data is finalised two to three days after the fact, and the most recent day in any report is often partial. A partial day is always a drop. QueryWatch labels the last complete day it is showing; in the web interface, assume the last day or two are incomplete and look at the one before.

Second, check what the drop is being compared with. A Sunday against a Saturday is not a drop, it is a weekend. The comparison that means something is the day against the same weekday in the weeks before it. In the demo, Sunday, Sep 6 had 412 clicks against a usual Sunday of 624, which is 34 percent down and well outside the normal spread. That is real. Move on.

Minutes 2 to 5: which pages?

Open the change investigation, or in the web interface, compare the day with the previous week in the Pages report and sort by the difference. What you want is not the list but its shape.

In the demo, three pages explain 88 percent of the drop: /pricing lost 118 clicks and accounts for 56 percent on its own, /features/automations lost 51 for 24 percent, and a blog post lost 18 for 8 percent. That concentration is the first real finding. A drop that lives in one or two pages is a page problem: something changed on the page, in how Google indexes it, or in what Google shows for its queries. A drop spread thinly across fifty pages is a site problem, an algorithm update, seasonality, or a tracking artefact, and the rest of this investigation looks different.

Here it is a page problem, and the page is /pricing.

Minutes 5 to 8: which queries, and what kind of loss?

Now look at the queries for the top page, and ask one question: did it lose impressions, or did it lose clicks while impressions held?

Fewer impressions means Google is showing the page less, or lower. Check the position for the main queries. If the position fell, the page lost ranking; if the position held but impressions fell, demand fell or Google is showing the page for fewer queries.

Clicks down with impressions steady is a click-through problem, and it almost always has a visible cause: a title Google has rewritten, a competitor’s rich result sitting above you, a new SERP feature pushing the result down the page, or a date in the snippet that now looks stale. These are the drops that recover fastest, because the fix is on your side.

Minutes 8 to 10: the counter-mover

Look for the page that moved the other way on the same day. QueryWatch shows it next to the contributors; in the web interface, sort the comparison the other way.

In the demo, a blog post gained 22 clicks while /pricing lost 118. The gain is small, but what matters is whether the two pages share queries. If a page lost clicks and a sibling page gained them for the same queries, Google has swapped which of your pages it shows, which is a cannibalisation problem, and the fix is on your site rather than in the SERP. If the counter-mover is unrelated, it is just noise in the other direction, and you can ignore it. Spotting cannibalisation has the full method.

Minutes 10 to 12: country and device

Split the drop by country and by device. A drop confined to one country points at something local: a regional competitor, a localised SERP change, an outage in a CDN region. A drop confined to mobile with desktop holding steady is a rendering or page experience issue, or a mobile-specific SERP feature. A drop that is uniform across both says the cause is upstream of either, and that is the more common and less fixable case.

Minutes 12 to 14: is the page still indexed?

Run URL inspection on the top contributor. You are checking three things: that the page is still indexed, that the canonical Google chose is the one you expect, and when it was last crawled. A page that dropped out of the index, or whose canonical quietly moved to a parameter version of itself, explains an overnight drop entirely, and is the kind of thing a deploy breaks without anyone noticing. The free tier covers five inspections a day, which is enough for an investigation like this one.

While you are there, open Search Console’s messages. A manual action or a security issue will have an email too, but it takes ten seconds to rule out.

Minutes 14 to 15: write the sentence

The point of all this is to email the client before the client emails you, and the email is one paragraph: what moved, by how much, the likely cause, the next step, and when they will hear from you again.

Clicks on brightloom.co were down 34% on Sep 6 against the previous week. Most of it is /pricing, which is still indexed and still ranking but is getting fewer impressions for its main queries. I’m checking whether a competitor result moved above us and will update you Friday.

That is the entire product thesis, and it works without the app too: the difference between explaining a drop and being asked about one.

What you cannot see from your phone

Be honest about the limits. Server logs, how the page renders, what the SERP actually looks like for the query in the client’s country, and whether a change shipped on Tuesday all live on a laptop. The fifteen minutes are for finding out which of those to open first, and for getting your sentence in before theirs.

If you would rather the first two minutes happened without you, that is what the alerts are for: the notification arrives already pointing at the page.