Is the drop real? Check the data first

A fall in organic traffic arrives with an assumption attached: something is wrong with the site. Sometimes it is. First, though, it has to be established that the drop is real, in the sense that the data is sound, the calendar and demand do not explain it, and the fall has been located in the report. Our overview, diagnosing a traffic drop before acting on it, lists the causes Google describes. This article stays with the first of them. We do not repeat the definitions in our piece on what the Performance report can and cannot tell you; this one is about the order of the checks, and how to leave a record others can verify.

Settle the dates and the unit first

Many disagreements about a drop are really disagreements about two different date ranges. Google’s Performance report overview describes settings that change what a chart shows. Granularity can be hourly, daily, weekly or monthly; weekly and monthly views smooth out effects such as weekends and holidays, while daily helps to spot spikes. The newest data can be preliminary, marked with a dotted line, and may change within hours.

Google’s page on troubleshooting data discrepancies adds that collected data is usually available in two to three days, and that outside the 24-hour view daily data is labelled in Pacific Time, so other systems on a different time zone may not match day by day. A dip in the last two days may therefore be incomplete data. Before comparing anything, agree the date range, the granularity and the comparison period.

Test the calendar and the demand

Google’s guide to debugging traffic drops says to widen the date range to 16 months, so the drop is seen in context, then to compare the last three months with the previous period or with the same period a year earlier. Its example chart shows a recent drop that happened in exactly the same way the year before.

For demand, the guide suggests filtering to your biggest queries one at a time and checking each in Google Trends, to see whether the dip is specific to your site or happens across the web. Google’s Trends data FAQ sets the limits of that check. The data is a sample, normalised on a 0 to 100 scale, low-volume terms appear as zero, and it is not a perfect mirror of search activity. Google says it should be considered as one data point among others. In our reading, that makes Trends good at supporting a claim that demand moved and poor at proving the site is not at fault.

Slice the drop one dimension at a time

The debugging guide says to click through every tab and check whether the change is confined to particular queries, pages, countries, devices or search appearances, and to analyse the search types separately. The Performance report’s search type filter now separates web into text-based and multimodal (searches that start from an image, such as those from Google Lens), alongside image, video and news. Per Google’s dimensions and data groupings page, the queries dimension is not available for multimodal traffic, because there is no text query to report.

The same page describes a branded and non-branded query filter, with 16 months of history from its introduction in March 2025. It is not available to sites with few impressions, and Google says some queries may be misclassified. The compare function evaluates two devices, countries, search types or search appearances side by side. In our reading, a fall confined to one country and one device suggests a narrower cause than a fall across every slice, and a fall that appears only in brand queries is partly a question about brand demand, not about the site.

Slices will not always add up. Google’s troubleshooting page says the table shows at most 1,000 rows, and that anonymised queries are in the chart totals unless a query filter is applied.

What impressions and clicks moving apart can mean

The shape matters as much as the size. Google says that if both impressions and clicks dropped, you work through its list of common causes. If impressions held and clicks dropped, the title and snippet may not be helping users understand the page, or other sites had a more appealing rich result. A small shift in position can also produce a noticeable traffic drop without a big change in impressions. Google adds that absolute position should not be the focus, though a dramatic, persistent drop in position is the cue to self-assess content.

The search appearance slice can expose changes outside the site entirely. Google’s data anomalies page records, for example, that FAQ rich results stopped appearing in Google Search from 7 May 2026, which reduced reported FAQ impressions. A drop in that appearance type, with nothing else moving, has an explanation that no change to the site would fix.

When Analytics and Search Console disagree

Google’s guide to using Search Console and Google Analytics data says the two tools use different metrics and systems, so the numbers will not match completely. It recommends comparing Search Console clicks with Analytics sessions, using the same country and device filters on both, and says that what matters is whether the general trends follow the same pattern. A small difference can be ignored; a big one is worth investigating. We cover the full list of reasons in using the two tools together.

Google notes that Analytics depends on a tag, so a missing tag on some pages affects its data, while Search Console data is processed uniformly. In our reading, that sets the order of work: if Search Console clicks are steady while Analytics organic sessions fell, check the tag, consent handling and channel grouping before looking at search. If both fell together, the drop is more likely to be real, though other causes remain open.

Recording the baseline so the result can be checked

Nothing in Google’s documentation describes how to write this up, so what follows is our reading of what makes a finding checkable. Record the date range and comparison period, the granularity, the metric, the search type and every filter applied, then what the anomalies page showed on the day you looked and what Trends showed for the queries you tried. A baseline is only useful if someone else can repeat it and get the same chart.

Where this fits at Cultured Digital: ruling out before diagnosing

Our way of working starts with investigating, then diagnosing the real problems rather than treating symptoms. Ruling out measurement, calendar and demand is part of that, because it decides whether there is a problem to diagnose. Our SEO audit is a focused investigation that finds the cause, then says what to do first, and it can take in the technical areas covered under Technical SEO. Where the question is how measurement is set up, the Measure guide covers measurement design and reporting that answers business questions, and Understand the market covers search opportunity analysis and keyword research. If the data holds up, the next questions are technical causes, a Google update or a site move or release. We do not promise rankings or traffic numbers; we set out the evidence and measure against it.

Is the drop real? Checking anomalies, seasonality and reporting first