The Importance of Data Quality in Marketing Attribution

One thing I always tell brands is that Facebook Retargeting and Facebook Prospecting are really two different channels. Your retargeting campaigns are likely to have low incrementality, whereas your prospecting campaigns might actually be undercounting, depending on what you’re using for attribution. Attribution expert Mike Taylor discusses this in more detail:

I was struck by an interesting example of the importance of data quality in talking to the Measured team recently. They showed me a slide that demonstrated the enormous difference between tactic- and channel-level incrementality. 

Here it is:

This chart shows the platform reported ROAS (Return On Ad Spend) vs Incremental performance across the Measured portfolio of brands, at the tactic level. This is the difference between what the platform reported vs what Measured found via incremental lift testing. If Meta reported $2.56 ROAS for a Facebook campaign, the true figure, as proven by turning spend off in some geo-graphic regions and not others, was $1.74.

Tactic-Level vs Channel-Level Incrementality

This is the bread-and-butter of what Measured do, but what’s interesting here to me isn’t just that they’re able to prove incrementality, but how radically different the numbers were between tactics within the same channel. On average, Facebook retargeting reported $5.30 ROAS, more than double what was reported for prospecting, and yet the true incremental figure was the same at around $1.75. This happens because you’re retargeting people who were already likely to buy, because they’re on your email list or browsing your website, and many of them would have bought anyway without the ad.

There’s an even more egregious case in Google Ads; for example, non-branded search (also called generic terms), were driving a real ROAS of $1.69, very close to the $1.79 reported in the Google Ads platform. However, the branded search terms are reporting a whopping $11.20, when in reality, the true incremental ROAS was $2.20. We know brand search is valuable (particularly when competitors are bidding), but many of those customers would just click on your organic listing if you stopped running ads.

The Importance of Data Quality

This is where data quality comes in. If you simply bucketed all of Facebook or all of Google together as one channel, you would have missed these disparities and would be wasting your advertising budget. Behind the scenes, to make this report a reality, Measured has found ways of programmatically breaking out each channel into individual tactics, like prospecting and retargeting. Then they run incrementality tests at the tactic level, in order to find these big differences within the same channels.

This is the hardest part of building a benchmarking tool because so many ad accounts are poorly labeled and use different naming conventions. You can’t just rely on if the campaign has ‘prospecting’ in the name because some people call it ‘broad’, ‘lookalike’, or ‘TOFU’ (Top Of Funnel). I’ve seen ‘retargeting’ called ‘remarketing’ or ‘BOFU’ (Bottom Of Funnel), and a hundred different permutations. Even if the naming is solid, the actual targeting options might not match the name; for example, somebody forgot to exclude existing customers, as was shared in a recent case study.

On top of the complexity of breaking channels out into individual tactics, is being able to harmonize the definition of ROAS across multiple clients. You might think that DTC E-Commerce businesses are largely similar, but many of them use multiple conversion events to report revenue. In some cases, indicator metrics like ‘add to cart’ will have a value attached (which isn’t real revenue), and some retailers are using server-side tracking events to capture revenue. The Measured team has been working for over 7 years to perfect these systems.

Conclusion

The takeaway for your business is that you can’t treat every channel as homogenous. There are real differences to different tactics, to the point where Facebook Retargeting behaves closer to Retargeting on other channels, than it does to Facebook Prospecting. Data quality and naming conventions are key to finding these under-performing segments, and you must pair that with proper incrementality testing at the tactic (not channel) level, to really understand your true marketing performance.

To learn more about how Measured can help your brand, schedule a demo today.