Search demand vs buying intent: the two curves behind seasonality

On a trip-planning subscription we manage, searches held steady through the summer while the share of sign-ups that bought dipped only in the travel months. Cost per sale rose in summer, and the cause was not fewer searches. That is one account, so it carries no figures here. It is the reason we now read seasonality as two curves rather than one.

The first curve is demand: how many people search for the product each month. The second is intent: of the people who arrive and sign up, how many go on to buy. A cost per sale can rise because the first curve fell, because the second fell, or both, and the word "seasonal" hides which.

This is our arithmetic, not a platform rule. For a product sold after a sign-up, cost per sale is cost per sign-up divided by the share of sign-ups that buy. Demand works mostly on the first half, through how many people search and what a click costs. Intent works on the second half, after the ad has done its job.

So the same rise in cost per sale can come from fewer searches or from the same number of searches turning into fewer buyers. The total moves the same way in both cases. Only the split says which one you are looking at.

For demand we use Keyword Planner's Avg. monthly searches. Google defines it as the average number of times that people have searched for a keyword and its close variants based on the month range as well as the location and Search Network settings that you selected, and says you can use this information to see how popular your keywords are during different times of the year. By default, Google says, it is averaged over a 12-month period. Our practice is to read it month by month rather than as the headline average.

Two cautions from the same page. Historical stats like average monthly searches are only shown for exact matches, and search volume statistics are rounded. And the forecasts beside it are a different thing: Google says forecasts take bid, budget, seasonality and other factors into account, while historical metrics don't. We read demand from the history, not the forecast.

In our accounts, intent mostly does not live in Google Ads for a product sold after a sign-up. Google says conversion rate ('Conv. rate') is calculated by dividing 'Conversions' by the total eligible interactions (e.g. ad clicks or video ad views). That is a rate on the ad interaction. The share of sign-ups that buy happens later, in our accounts often weeks later, and we take it from the product's own records rather than from the ad platform.

Our practice: count sign-ups by the month they signed up, count which of them bought, and only read a month once enough time has passed for its buyers to have bought. A recent month can look worse before it finishes. Google describes the same effect for ad conversions on its conversion delay page: if you compare recent performance with past performance, your recent performance might not look as strong, because some of the people who clicked your ad haven't converted yet.

This is our reading of the account, not something Google states. People plan a trip before they take it, so searches for a planning tool can hold up through the summer. But a share of the people who sign up in the travel months are already away or about to be, and they buy less in that month than the same number of sign-ups would at another time of year. Demand did not fall. Intent did.

A product with a different buying cycle will have a different pair of curves. The point is not the shape for trips. It is that the two curves are measured separately, so each has to be read before either one is blamed.

Our reasoning, not a platform rule: when demand falls, there are fewer people to reach, and a higher bid mostly buys a bigger share of a smaller pool. When intent falls, the people are still arriving, and the question is what happens after they sign up: the offer, the timing, the follow-up. Those are owned by different people, which is the practical reason to split them.

On our reading, the nearest Google tool works on a different rate. Google describes seasonality adjustments as an advanced tool that can be used to inform Smart Bidding of expected changes in conversion rates for future events like promotions or sales, and says to use them only if you expect major changes to conversion rates, because Smart Bidding already manages seasonal events. It also describes them as ideal for short events of 1 to 7 days, and says they may not work as well for extended periods of more than 14 days at a time. Our reading: where the tracked conversion is the sign-up, that is the ad conversion rate, not the share of sign-ups that buy, and a dip that lasts the whole travel season is not what the tool is built for.

This is our practice, drawn from too few accounts to carry a number. Before a month's cost per sale is called seasonal, we read searches for it from Keyword Planner and the share of its sign-ups that bought from the product's records, and say which of the two moved. If searches moved, the plan is about budget and reach. If only the buying share moved, the plan is about what happens after the sign-up, and the ads are left alone.