Social media reporting: These are the KPIs you need to watch

The lack of measurability in traditional advertising has led to almost blind faith in the field of performance marketing. People are so happy about hard revenue data that they often don't really question the foundation. We have summarised for you what you need to pay attention to so that the data is truly valid.

First things first - Transparency

The foundation for every successful collaboration is transparency. A sign of any good agency is that it has no problem providing access to the respective advertising account. Reporting without the ability to check the raw data can be considered pure fiction. We have already uncovered too much wrongdoing during audits for clients.

Complete transparency creates a common honest basis where no one can hide behind PowerPoint slides, but only the shared, real success counts. That is why we recommend that clients in all advertising areas only work with partners who grant transparency and do not try to hide anything.

Incremental purchases

Not every purchase that is attributed to an advertising measure is actually incremental. This means that the purchase would not have occurred anyway (even without advertising) at the same or a later time. Offline, this is an issue with large discount campaigns, for example, where science shows that many of these purchases would have been made at a later time without a promotion (and at the regular price).

Just because someone sees an advertisement, clicks on it, and buys, it does not mean that they would not have bought it otherwise. Thankfully, there are ways to find out what proportion of purchases this applies to.

1. Conversion Lift Studies

With Conversion Lift studies on Facebook, the campaign target group is segmented. A control group does not see any advertising during the investigation period. Afterwards, the proportion of buyers between the groups is examined.

Surprising results often come to light here: from the worst-case scenario that hardly any incremental purchases were achieved, to the best case, when it is shown that even pure ad impressions (without clicking on the advertisement) led to incremental purchases. Conversion lift studies require a minimum budget, which varies depending on the country and can be enquired about with the respective Facebook Partner Manager.

‍2. A/B test with holdout

The same principle is behind this methodology, but it can also be implemented with smaller budgets. Again, the target group is segmented and a control group is defined that does not see any advertisements. The statistical significance of this method is lower than with a conversion lift study, but you still get a good overview of the effectiveness of the advertising measures.

Attribution windows

Every platform measures purchases differently. The most important feature is the attribution window, which in turn can be separated into view and click attribution. With view attribution, all purchases are attributed to the advertisement for which the ad was previously seen (but not necessarily clicked). With click attribution, only those purchases are attributed where a click on the advertisement also took place. However, this does not mean that the click led immediately to the shop, shopping cart, and checkout. The purchase can also happen subsequently in terms of time.

This is where the attribution window comes into play, which defines the period of how long a purchase is still credited after seeing or clicking on an advertisement. The platforms have different windows to choose from here. On Facebook, for example, you can choose between 1, 7, and 28 days.

A short window is not necessarily correct, because for many products, consumers often think longer before they buy. Travel, for example, is a case where the decision period experience has shown to be around 14 days.

However, as a client, you should always have the window confirmed in writing (and check it), as well as regularly look at the breakdown. If, for example, hardly any purchases have occurred after clicks and the purchases after views are evenly distributed over 28 days, this indicates that hardly any incremental purchases were generated.

Tip: Work across platforms with attribution models that can distribute a purchase across multiple advertising contacts; more on this in a future article.

Retargeting is a beast

There is no successful performance marketing without retargeting, but there is also hardly anything that is so often used and reported abusively to make performance marketers look good.

It is simply easier and cheaper to persuade someone to buy who has already been to the website and viewed a product than someone who does not even know the brand yet. The problem with this: someone who has already been on the website would with a certain probability have bought a product even without additional advertising delivery via retargeting.

Therefore, one should always pay attention to what proportion of purchases comes via retargeting or the real incremental value is measured via conversion lift studies or A/B tests with a holdout.

You should look closer if almost all purchases come via retargeting and almost nothing via prospecting (addressing new target groups). In individual setups, this may be intended, but it is definitely noticeable. If you are not sure, you should ask the agency to show both values separately in the reporting.

Additional tip Google

We are currently rolling out a high-end Google Ads programme exclusively for existing customers. In this area, the following classic case often comes to light during audits: purchases that arise after searching for one's own brand terms are almost never incremental. For a brand that thought it was successful on Google, we found out that almost all purchases came via such terms and thus hardly any additional purchase was ever generated.

Because Google functions more to meet demand (unlike social media, which is more demand-generating), you have to be doubly careful. Typically, the effect of Google is overestimated because it is the last step in the customer journey. Budgets are thus disproportionately allocated to a channel that brings fewer incremental purchases in proportion. However, this of course depends very much on the shop in question and must be evaluated individually in each case.

Conclusion

Performance marketing is an essential part of every marketing strategy, but just because something is measurable does not mean that the right thing is being measured.

For clients, it is essential to build up an understanding themselves in order to be able to evaluate any service providers and make the right budget decisions. Because the budget distribution should be designed in such a way that those channels that bring incremental purchases and thus provide additional revenue are taken into greater account.

Conclusion