In recent months, the keyword “big data” was the Holy Grail of sorts in the marketing realm. But until now, many discussions around the topic were primarily carried out at congresses and conferences in the more theoretical form of the Knights of the Round Table. This is especially so when it comes to external data (third-party data) which advertisers purchase in addition to their own data (first-party data), in order to control more targeted online campaigns.
This is because up until now, campaigns with third-party data only existed in the specialist lectures of most international advertising service providers, but unfortunately, had far too little presence in the German online advertising market. The infrastructure stemming from ad servers, data management, and demand-side platforms was available, but data suppliers were missing who could help a market to get off the ground.
But this situation is changing and more and more companies are also offering data for sale which is relevant to the German market. This provides a good reason for advertisers and their service providers to ask two central questions in particular:
- How much uplift is third-party data expected to supply to my campaign?
- With that in mind, how much might third-party data cost, in order to ensure that the campaign remains at least as efficient as it was before — and in an ideal case, even more efficiently?
The answer to the first question in particular is a difficult one, because, for one thing, “the ultimate campaign” does not exist. By the same token, as a general rule, advertisers have little to no empirical knowledge regarding the use of third-party data.
Therefore, I recommend that the question be asked differently and restated in terms of the second point below: If third-party data costs a specific amount, then how high must the uplift of the campaign be, in order that efficiency remains at least at the current level? And if the result is that there must be a minimum uplift of over 30%, then at such a point in time at the latest, an additional purchase of external data should be more closely scrutinised or all media alarm bells should ring.
In order not to surrender in advance, I thereby offer—without obligation and free of charge—my three rules of thumb which can be helpful in the use of third-party data in digital marketing.
1. Examine data quality very carefully!
Data offered from a supplier or data management platform must absolutely be put to the acid test before purchase. Enquire about how the data is labelled and if it really comes from the market in which it is to be later used.
Pay attention to how the data was generated: Is it “hard” data, or were projection algorithms used in the generation of data? If the data was originally collected in the offline world, it must also be examined as to whether the matching procedures conformed to data protection laws. And, last but not least, the question arises as to whether the specified quantity and granularity of the data profile is truly credible in relation to the total size of the target audience.
It is essential to consider in advance how you can analyse the quality of the purchased profiles. For example, measure the hit rate of the purported characteristics, e.g. via a panel or an online survey. Or, alternately, are there other measurable key values (KPIs) in the campaign which are to be improved by the data? If the answer to both questions is “no”, then steer clear of this data.
2. Choose the shortest path!
Campaigns that rely on third-party data can run into a quantity problem very easily. Why? Because the quantity of available profiles are generally less than the desired amount. This is particularly the case if the target audience is especially narrow and, at the same time, the data quality is expected to be high.
In order to explain why the quantity of available data is so important, a small technical digression is unavoidable: In the use of external data in a campaign, unfortunately, all acquired cookies are never obtained. This means: Some data sets are purchased, but cannot be used. This occurs, for example, when a portion of the cookies have been deleted by the users since then, or originate from another group of users, who are not in the environments in which attempts are made to find these users again.
This shrinkage is exacerbated by the fact that during the transfer of data from the supplier’s system to the buyer’s system, a synchronisation of cookies must take place over the user’s browser. Both systems must exchange their cookie IDs as well. The quantity of data is reduced considerably by means of this cookie synchronisation, because at some point, every user must be found on the website by the system for this purpose. Our experience shows: Even in the best-case scenario, about one-fifth of the available profiles are lost during this process. If the cookie synchronisation is poorly implemented, then very quickly, more than half of the profiles can be lost.
The risk of loss can be minimised by allowing the cookie synchronisation to occur as near as possible to the place that the data was generated or put online. Thus, all unnecessary partners within the supply chain are eliminated! These only cost money and reduce the quantity of usable data. In case the data supplier already employs a data management platform, the data can possibly be transferred directly into the delivery system, and in this way, you avoid an additional data synchronisation process.
3. Recalculate in advance!
A simple calculation can decide the fate of your data campaign: Set the price that you should pay for third-party data in accordance with the added value that you must achieve by the use of external data. Do data costs eat up the performance improvement that the campaign is meant to achieve? In this case, the use of third-party data would not improve the campaign’s efficiency. If you still have no empirical knowledge as to whether the uplift that must be generated by the data is realistic, then ask experts who can present you with benchmark variables.
In the purchasing of third-party data, take profile quantities and target audience sizes into account. The acquisition of data does not always pay off. Thousands of users that were, for example, identified as clear-cut interested parties for an especially strong-smelling type of stockfish might be a valuable target audience. However, it is rather unlikely that it is worthwhile to address these people though a narrowly focused re-targeting with a display campaign. Here, it would be better to search for other, more cost-efficient methods of reaching these fish aficionados.
Whether the use of external data in online marketing is worthwhile can generally only be assessed by companies in retrospect. However, it does not hurt to carry out a few simple calculations in advance. At any rate, the hardest currency is the experience gleaned with the performance values of campaigns. Wherever suppliers of third-party data can provide satisfaction—both when it comes to the quality of data as well as the price—then they have a rather good chance to belong to the Round Table of advertisers in the future.
First published in German by internetworld.de.