A lot of businesses see a lot of seasonal variation in their sales, whether it’s a toy retailer at Christmas, or a holiday firm in the middle of the summer. Car retailers may see a spike in traffic when new registrations come out, and debt-help and loan websites are very busy in January and February.
Generally, these periods are good for businesses, but it can be the case that the performance doesn’t quite match the increase in traffic. Too often, businesses simply label the visitors as lower-quality traffic, and adjust their bids accordingly, but this can be totally wrong.
Suppose that a business has a significant lead time between a visitor hitting their website for the first time, and the time of the sale – more than a week, perhaps. When traffic volumes are stable, so are the conversions, but what happens when you add seasonality to the mix?
Here’s some (obviously) fictional traffic data to a website that sees a spike in its clicks in the middle of summer:
Imagine first that when somebody makes a purchase, they always do so on the same day. For the sake of an easy-to-read graph, we’ll say they have an unlikely 20% conversion rate:
So far, no problems. The conversions increase at the same time as the clicks, and the conversion rate remains constant, at 20%.
Now consider a business where people take more time to consider the alternatives before converting.
What happens to the above example if you still have a 20% conversion rate, but 5% convert one week after the initial visit, 10% after two weeks, and 5% after three weeks?
Suddenly, the conversion rate is all over the place, dropping without any obvious reason in the early spring, then increasing through until late autumn. What’s caused this?
Quite simply, if you see an increase in your clicks, but the conversions don’t come until later, your conversion rate will fall. Put another way, your conversions are still being driven from past clicks (which were lower).
In the above case, I’ve used a fairly improbable sinusoidal curve for sales. What happens if sales are usually flat, but there’s a sudden, sharp increase?
I’ll use the conversion-delay distribution from before, with the following click distribution:
The variation in conversion rate caused by this seasonality is huge:
The conversion rate drops from 20% to 15% immediately, the moment the clicks start to increase. It starts to recover, but the moment the clicks start to fall, the conversion rate goes through the roof.
The Catch
Looking at the above graph, it’s fairly clear what’s happened. But understanding what’s happened in real time from looking at the data isn’t all that easy. Look at the performance data from the above graph up to week 21:
What are you going to conclude at this point? A lot of analysts, even quite experienced ones, would assume that the additional traffic to the site is of lower quality, and will cut the bids.
Of course, this is completely the wrong thing to do – the clicks aren’t worth any less.
Similarly, when the conversion rate increases, there’s a temptation to increase the bids:
Again, this is the wrong move – the clicks are worth the same in reality, so the bids shouldn’t change.
The Bottom Line
Individually, seasonality and delays between clicks and conversions don’t really do much harm. Combining them, you see some eye-opening variations to your conversion rate, which can lead you to make some very bad decisions.
This is also a problem if you decide to significantly change your budget and bids. Suddenly driving more or less traffic to your site can make a big change to your conversion rate. Have you ever seen that your conversion rate drops when you increase the bids, or climbs when you cut back? It may not be a reflection of the position of your adverts, but a natural result of the lead time on your conversions.
As long as you know that this is a possibility, you should be able to respond in a more measured way, but how many PPC Analysts look at the delay between the click and the conversion on a campaign? I would strongly urge anyone with seasonal demand or long delays between first click and conversion to check this out before their next peak or dip in business.