Changes in seasons can trigger changes in the amount of searches that people use for certain queries and topics.
For many websites owners, understanding those seasonal variations may lead to more visits from people who are interested what they offer upon their pages.
By seasons, I don’t just mean winter, spring, summer and fall. There are many different seasons that can affect how and what people search for, such as seasons for baseball, football, basketball, and hockey.
Or seasonal variations based upon recurring holidays such as Christmas, Valentines Day, Mothers Day, and Thanksgiving. The start of a school year can trigger certain searches, and summer break from school can impact other searches.
If you own an ecommerce business, you are probably familiar with the value of creating content and offering products that take advantage of periodic changes that happen at different times of the year. More informational sites, such as online newspapers or magazines, blogs, can also benefit from an awareness of patterns which people use in searches.
One of the difficulties of planning for seasonal traffic has been with the tools available to try to research the amount of volume that different search terms might possibly bring to a website, or at least allow someone researching those words to compare different choices that they might use. Microsoft unveiled a beta keyword research project a little over a year ago – the Search Volume Seasonality Forecast – which showed seasonal trends for keyword usage.
Microsoft’s program is still in Beta, and the searches you can perform are very limited. It’s difficult to tell if they will ever expand the tool, but it’s possible that they might. Microsoft had a patent application published that appears to describe that tool, and how it might work:
Invented by Lee Wang, Li Li, Shuzhen Nong, and Ying Li
Assigned to Microsoft
US Patent Application 20070239703
Published October 11, 2007
Filed: March 31, 2006
A method and system are provided for forecasting keyword search volume. Keywords are categorized by concept and by the amount of data available for use in predicting future behavior. The keywords and/or the categories can also be categorized as seasonal or non-seasonal.
A category level seasonal variation pattern can then be calculated based on keywords in the category that have sufficient historical data. A search volume can then be predicted for one or more keywords, with an appropriate calculation algorithm being selected based on the concept category, seasonal classification, and historical data available for the keywords.
The patent application describes a way to generate forecasts of keyword search volume. While this kind of forecasting can be useful for paid search, it’s not unusual for search marketers to also look at search volume information when looking for keywords to use for search engine optimization.
It appears that some of the technology behind the methods are described in a Microsoft paper that shares a number of authors with the patent filing – Similarity of Temporal Query Logs Based on ARIMA Model.
For some keyword phrases, where there is a couple of years worth of data on searches using those terms, it may be possible to classify them as seasonal or non-seasonal terms using a statistical model like the one described in that paper.
For keyword phrases where such data doesn’t exist, it might be possible to see if those phrases fit into concept categories with other phrases where there is data regarding how influenced they are by other seasons. If a phrase can’t be classified in that manner, a seasonal forecasting method might fall back on looking at the search volume for the term for the previous month.
There’s no telling if or when the Microsoft Seasonality Forecasting tool might come out of Beta, but I’d like to see it released. Perhaps now that the patent application has been published, we might see it move out of the Microsoft Adcenter Labs sometime soon.
There are some keyword research tools, like Keyword Discovery, that do provide a years worth of Seasonal Trends, though it would be nice to have more than a single years worth of data to look at. Being able to identify similar phrases that may also be seasonal (but don’t have a history of data to use to tell) would also be a good option, if it worked as well as the Microsoft patent application seems to suggest.