Seasonality: why traffic moves without any change from you

Demand for most products rises and falls on a yearly pattern, so a change in traffic is not always a change in your results.

Search traffic does not stay flat. People look for tax help in spring, heating repairs in autumn, gifts before holidays and gym memberships in January. Those swings come from how many people want what you sell, not from anything you did. Mistaking a seasonal move for the effect of your work is one of the most common errors in reading results.

What seasonality looks like#

• Yearly cycles: a peak at the same time each year.
• Weekly cycles: business to business sites often fall at weekends; shopping sites often rise.
• Event spikes: a trade show, a holiday, a school term or a deadline.
• Slow dips: summer and the end of year holidays reduce searches in many fields.

How to check for it#

1. In Search Console, set the date range to the last 16 months and compare it with the same months a year earlier. If last March and this March look alike, the shape is seasonal.
2. Open Google Trends and type in the main phrase your customers use. It shows the yearly pattern of interest in that search, even without any of your own data.
3. Compare year on year, not only month on month. A fall from December to January means little if it falls every year.
4. Compare impressions as well as clicks. If impressions rose and fell with the season, demand moved, not your appeal.

What to do about it#

• Judge a change against the same period last year, or against pages that you did not touch.
• Plan ahead. If demand peaks in spring, update the page in winter, so it is ready and indexed when searches climb.
• Keep a short note of the dates of changes, campaigns and holidays next to your charts.
• When you test something, avoid starting right at a peak or a trough.

Common mistakes#

• Celebrating a rise that arrives every year.
• Panicking about a quiet month that is normal.
• Comparing a holiday week with an ordinary one.
• Having under a year of data and assuming there is no pattern.

This is the main reason a fair test needs a baseline. See how to run a simple before-and-after test, and how a change is measured for how PageSoar handles it.