What is seasonality? Seasonality is a set of predictable market changes that you can apply from previous years. It’s useful to this because you can use it to plan for the near future. For example, when the weather is warmer, people buy more ice cream – that’s seasonality. The equivalent holiday version of this example would be: the weather is warmer, and every household is buying ice cream for every other member of their household and all their relatives, so we and all our competitors shout all day and all night about ice cream, and have a special day when we give between 10% and 50% price discount on our ice creams – therefore, people buy more ice cream. This is not seasonality, it’s a complete snow-storm of buying and selling signals.
In the holiday period, which can account for 40% to 70% of a client’s annual sales, it seems to be the deal opportunity for retailers to factor in their increased advertising investment to gain additional attribution insights. Logic tells us that the greater number of impressions and sales will provide a more statistically robust understanding of the tactics that are particularly effective at driving sales all the way through the funnel, identifying smarter ways of allocating budget within and between channels, and spotting optimisations of ad groups, creative and placements.
But, as illustrated in the ice cream example, over the holiday period, these models go a bit haywire. Within all large companies, both brand and e-commerce marketers run campaigns unlike anything else they do all year. Black Friday promotions provide discounts and incentives that are wildly generous. This represents something far more than mere ‘seasonality’. In statistical terms, this is an annual anomaly.
As marketers, we have to appreciate that there is no way that an attribution model can map the conversion paths in the holiday period – which is full of branding noise, random tactics and market stimulation – with the trading conditions of the rest of the year. We also have to factor in the commercial imperative that the holidays are vital to many companies’ financial health. This means that companies have to spend money to compete in this crowded market, even if they are more prudent and scientific in their advertising spending decisions for the rest of the year.
So, in this mess of buying and selling signals, what can the savvy digital marketer do?
Assessing the effectiveness
Firstly, it is recommended that you focus on extracting the maximum insight for the next time you face these unique trading conditions – namely, next year’s holiday madness. To do this, it is not practical to develop a separate attribution model that embraces the different conditions and tactics that are in play in the holidays. But, it is advisable for you to put in place a testing framework that will allow you to assess the effectiveness of your holiday tactics – i.e. varieties of weights and deployments of paid social, video, incrementality of affiliates and search testing – which you should be changing dramatically from business-as-usual ad spends at other times of the year.
The reasons for this change of tactics is self-evident: this is a time of year when you will be opening budgets to maximise sales, and efficiency will be a less important filter on your spending decisions. The significant changes that you’ll make to channel and ad group configurations – most notably up weighting display and opening your search budgets – should be accompanied by tests to understand the different customer journey times so that you can make better reach and frequency decisions next year.
At this time of year, massive awareness campaigns raise demand throughout your market, and this change in the consumer environment – alongside the discounts and bundle offers that you may be offering – will require a new set of metrics that your attribution toolset can help you bring to the surface.
Isolate your findings
Secondly, these findings need to be isolated from the attribution findings that you apply in the other months of the year. Any learnings from the holiday period should be applied as far as possible only to those unique market conditions. For example, the length of the conversion path in December should not be applied to your planning in any other month of the year.
Even if your investment levels in brand and awareness are as high as they are in December, all of the other special conditions of the holiday period will not be in place. We recommend that your review your look-back window when you are running attribution reports in February and March so that you can minimize the skew that the December buying frenzy will have on converging paths.
The holidays create a more compelling shopping event than any other in the UK, and in attribution terms, these are unique testing conditions. That’s just how it is – you can’t have everything at Christmas.