A page that used to bring in steady traffic is getting fewer clicks. Before you rewrite anything, find out which kind of drop it is. Search Console can usually tell you, because the four numbers it reports for a page (clicks, impressions, click-through rate and average position) move in different combinations depending on the cause.
This guide walks through the checks in the order we'd run them. You need access to the site's Search Console property and about twenty minutes.
Step 1: Confirm the drop and size it
Open Performance → Search results, click Date, choose Compare and pick Last 28 days against Previous period. Then add a filter for the page (+ Add filter → Page → URL is exactly) so every number below is for that page alone.
A few things keep this comparison honest:
- Compare whole weeks. Weekday and weekend traffic differ on most sites. A 28-day window contains four of each weekday, so both periods have the same mix. A 30-day window doesn't.
- Leave out the last few days. Search Console data usually arrives with a delay of two to three days, and the most recent days can still change. If the newest days look like a cliff, wait before you react.
- Check last year too. Search Console keeps 16 months of data. Compare the last three months with the same three months a year earlier. If the page dropped by a similar amount last year, you're probably looking at seasonality, not a problem.
Write down the size of the loss in clicks, not only as a percentage. A 50% drop on a page that had 30 clicks a month is a footnote; a 15% drop on your best page is the week's priority.
Step 2: Read the four numbers together
Turn on all four metrics at the top of the chart. Then compare how each one moved:
| Clicks | Impressions | CTR | Position | Most likely cause |
|---|---|---|---|---|
| Down | Down | About the same | Worse | The page ranks lower: a competitor, a content change on your side, or a ranking update |
| Down | Down | About the same | About the same | Fewer people search for it: seasonality or falling interest. Check the year-over-year view |
| Down | About the same | Down | About the same | The listing gets fewer clicks: a rewritten title or snippet, or new elements above it on the results page |
| Down sharply | Close to zero | n/a | n/a | Google stopped showing the page: an indexing problem. Go to Step 4 first |
| Down | Down | Up | Better | Often a reporting change rather than a real loss: see Step 6 |
Treat the table as a first guess, not a verdict. Average position is an average across every query the page appeared for, so losing a few low-ranking queries can make position look better while traffic falls.
Step 3: Find the queries that moved
With the page filter still on, open the Queries tab below the chart. Because you're comparing two periods, each metric has two columns. Sort by the clicks of the earlier period, then look down the list for queries whose clicks fell the most.
You'll usually see one of three patterns:
- One or two big queries account for most of the loss. Look at the search results for those queries yourself (in a private window). Has something new appeared above you: an AI Overview, a video carousel, a "People also ask" box, more ads? Has a competitor published a better page?
- Many small queries each lost a little. That points to something about the page as a whole: a ranking change, a technical issue or a change in demand.
- The queries changed. The page now appears for different, less relevant searches. That often follows a content rewrite that changed what the page is about.
Step 4: Rule out technical causes
If impressions collapsed, check that Google can still crawl, index and choose the page before you look at anything else. These problems are the most damaging and usually the quickest to fix.
- Inspect the URL. Paste the page's address into the URL Inspection bar at the top of Search Console. It tells you whether the page is indexed, which canonical Google selected, and when Google last crawled it. If the Google-selected canonical is a different URL, Google is indexing that one instead.
- Check the status code. The page should answer 200, directly. A redirect, a 404 or intermittent 5xx errors all reduce how often it's shown. Our redirect checker shows every hop.
- Check robots.txt. A new
Disallowrule can block a whole section by accident. Test the page's path with the robots.txt tester. - Check for noindex. Look for
<meta name="robots" content="noindex">in the HTML and for anX-Robots-Tag: noindexresponse header. Plugins and staging settings add these more often than people expect. - Check the canonical tag. A canonical that points to another URL (or to the
http://version) tells Google to index that address instead.
The Pages report (under Indexing) also lists URLs that dropped out of the index and the reason Google gives, such as "Excluded by 'noindex' tag" or "Soft 404".
Step 5: Find out what changed on the page
If the page is indexed and healthy, compare it with how it looked before the drop. Your CMS's revision history, your deploy log or an archived copy will tell you. The changes that most often explain a drop:
- The title or meta description changed. A shorter, vaguer title can cost clicks even at the same position. Preview it with the SERP preview tool and check the live tags with the meta tag checker.
- Content was removed. A section that answered a popular question, a table, a FAQ. If a query in Step 3 lost clicks, search the old version for the answer to it.
- Internal links changed. A redesign that removed the page from the main navigation or from related articles can lower how important it looks.
- The URL changed. If the page moved, every old link needs a single 301 redirect to the new address, and internal links should point to the new one directly.
If you find a change that lines up with the start of the drop, you have your main suspect.
Step 6: Check what was happening in Google
Some drops have nothing to do with the page.
- Ranking updates. Google lists confirmed core and spam updates, with dates, on the Google Search Status Dashboard (opens in a new tab). If your drop started within a few days of one and many pages on the site dropped at once, the update is a likely cause. Recovery then usually means improving the content, not fixing a tag.
- Reporting changes. Search Console marks known data problems with annotations on the chart. Changes in how results are counted can also move the numbers without any change in visitors. In September 2025, for example, Google stopped supporting the
num=100parameter that let tools load 100 results at once, and many sites reported lower desktop impressions and a better average position in Search Console with no change in clicks. - Seasonality. If the year-over-year comparison from Step 1 shows the same dip, there may be nothing to fix.
Step 7: Fix one thing, then measure it
Once you have a cause, make the change, and write down the date. Then wait. Search Console needs at least a few weeks of data after a change before a comparison means anything; we use 28 days, the same window as in Step 1.
When the time is up, compare the 28 days after the change with the 28 days before it, using the same page filter. Keep two limits in mind:
- A before/after comparison shows what happened, not why. Seasonality, updates and other changes on the site happen in the same weeks. If you changed three things at once, you can't tell which one worked.
- Small pages are noisy. A page with a handful of clicks a day can swing by 20% from one month to the next for no reason at all.
How Semhack does this for you
Semhack runs the first steps of this process every day on the Search Console properties you connect, using the read-only Search Console API:
- Drops. A page is flagged when it had at least 40 clicks in the previous 28 days, lost at least 20, and lost at least 30% of them in the last 28 days. Each flag shows clicks, impressions and average position for both periods, a daily clicks chart, and the queries that lost the most clicks.
- Slow declines. A separate rule catches pages that slide over four weeks (at least 15% lower, and falling in at least three of those weeks), which a 28-day comparison can miss.
- What changed on the page. If the page has been audited twice, Semhack shows which of its 40 checks changed between the two audits, such as a canonical that now points elsewhere or a meta description that disappeared.
- Did it work? When you mark an action done, Semhack compares the page's clicks in the 28 days after with the 28 days before, once that data exists.
It doesn't replace your judgement: it puts the numbers you'd otherwise pull by hand in one place, ranked by how many clicks are at stake.