Skip to the indexed answers
Digital Marketing Index

usdigitalmarketingindex.com SEO

How can a seasonal decline in search demand be separated from a ranking loss?

By checking whether the site’s share of the available searches changed, or only the number of searches. A seasonal decline means fewer people are searching: impressions and clicks fall together, position holds, and the same months fell in the same way in previous years. A ranking loss means the site is shown lower or less for searches that are still being made: position worsens, the site’s share of impressions for a fixed set of queries shrinks, and the pattern has no precedent in earlier years. The two can happen together, which is why each test below is run separately. Clicks alone cannot tell them apart. Position, impressions for a stable set of queries, a year-on-year comparison, and an outside measure of demand can.

Multi-year seasonality checks

Seasonality repeats. The first test is whether this decline has happened before at this time.

  1. In Search Console, set the date range to 16 months and view clicks and impressions by week or month.
  2. Compare the declining period with the same weeks one year earlier.
  3. Where older data exists, in exports, analytics, or business records, extend to earlier years.
Observation Suggests
The same months declined by a similar proportion last year Seasonal
Last year was flat or rising in these months Not seasonal, or a new pattern
The decline began on a specific date, sharply An event: a site change or a search update
The decline is gradual and matches the calendar Seasonal
Only this year's data exists Seasonality cannot be confirmed from the site's own history; use outside evidence

Year-on-year comparison is the view to start from. Comparing this month with last month in a seasonal business shows the season, not the performance. Comparing this month with the same month a year ago removes the season and leaves the change.

Account for calendar effects: the number of weekends in the month, public holidays that moved, and events tied to dates.

Market-wide comparison

The second test asks whether demand fell for everyone.

Google Trends shows relative search interest for a term over time. Its help explains what the numbers are: Trends “normalizes search data to make comparisons between terms easier”, dividing each data point by the total searches of the geography and time range it represents (Google Trends help). The result is an index, not a count of searches.

How to use it:

  1. Enter the main topic terms for the affected pages.
  2. Set the country and a range of several years.
  3. Look for a repeating annual shape, and for where the current period sits on it.
Trends shows And the site shows Reading
Interest falling in line with previous years Impressions falling by a similar proportion Seasonal demand
Interest steady Impressions falling Site-specific
Interest falling Impressions falling faster Both
Interest rising Impressions flat A relative loss hidden by the season

The last row is easy to overlook. A site can lose ground in its high season and still post rising traffic.

Other outside evidence:

Source Use
The business's own sales or enquiry history by month Confirms the season from an independent record
Industry or trade body seasonal data Where published
Paid search impression volumes for the same terms, if campaigns run Demand on the same queries, independent of organic ranking
Competitor visibility in a third-party tool Whether others fell on the same terms; an estimate, to be labeled as one

Trends has limits. It is normalized and sampled, low-volume terms return little, and the index for a term is relative to all searches, so it can fall while absolute searches hold if total search activity grew.

Stable query cohorts

The third test removes the effect of a changing query mix by following the same queries through time.

  1. Define a fixed set of queries: the non-branded terms that mattered in both periods.
  2. For that set, compare impressions, clicks, and position this period against the same period last year.
Metric for the fixed set Seasonal decline Ranking loss
Impressions Down Down, or steady with position worse
Average position Steady Worse
CTR Steady Down, following position
Clicks Down in proportion to impressions Down more than impressions
Share of queries on the first page Steady Lower

Position is the separating metric. Demand does not move where a page ranks. If position for the fixed set is unchanged and impressions are down, fewer searches were made. If position is worse, the site was shown lower for the searches that were made.

Two cautions. Search Console’s average position is an average over impressions, so keep the query set fixed, or the average will move with the mix. And impressions are counted only when the site appears, so a page that drops far enough stops registering impressions at all. Falling impressions with missing queries can be a ranking loss that looks like falling demand. Check whether priority queries are still present in the data.

Segment further where the answer is unclear:

  • by page: is the decline across all pages on the topic, or on a few?
  • by device and country: is it everywhere?
  • by branded and non-branded: branded demand has its own seasonality.

What evidence shows a true loss

An annotated trend comparison, with invented figures for a heating topic, comparing a spring month with the same month a year earlier.

Measure Same month last year This year Change Note
Clicks, all queries 8,000 5,600 Down 30% The alarm
Clicks, this month against the previous month Down 40% Seasonal: last year showed the same month-on-month drop
Trends index for main term 40 39 Flat Demand is where it was last year
Impressions, fixed query set 100,000 82,000 Down 18% More than demand explains
Average position, fixed set 5.1 7.4 Worse The site is shown lower
Priority queries on first page 18 of 20 13 of 20 Fewer

Reading: the month-on-month fall is the season, as it was last year. The year-on-year fall is not, because demand is flat against last year while position has worsened and five priority queries have left the first page. This is a ranking loss occurring inside a seasonal trough.

Decision table:

Year-on-year clicks Outside demand Position, fixed set Conclusion
Down Down similarly Steady Demand decline; not a site problem
Down Steady Worse Ranking loss
Down Down Worse Both; estimate each
Steady Up Worse Hidden loss
Down Steady Steady Look elsewhere: results page features taking clicks, a tracking problem, or lost pages
Up Up more Steady or worse Underperforming the market

Evidence that points to a true loss:

  • position worse for the same queries, on the same device and country;
  • priority queries gone from the data or moved off the first page;
  • the decline starting at a date that matches a site change or a known search update;
  • particular pages or templates affected while others on the same topic are not;
  • indexing problems in the page indexing report;
  • no matching decline in outside demand.

Evidence that points to demand:

  • the same shape in previous years;
  • outside measures falling in step;
  • position and first-page counts unchanged;
  • the decline spread evenly across pages and queries on the topic;
  • the business’s own sales history showing the same season.

Report the conclusion with its basis: which comparisons were made, over what dates, with which query set. Where the site has no history from earlier years, say that the seasonal reading rests on outside evidence alone.

Recognizing the seasonal shape after the fact is one task. Predicting it in advance, in time to prepare content before the peak, is another, covered in the guide to how to anticipate seasonal keyword demand before the yearly search peak.

The test: compare with the same period last year, compare with demand outside the site, and look at position for a fixed set of queries; a season moves the first two and leaves the third alone.

Link trail

Reader comments on this answer

Add your comment to this answer

Your email address will not be published. Required fields are marked *