In 2020, ISBA and the AOP and PwC delivered a ground-breaking study into open-web programmatic supply chains. In January 2023, they published a follow-up study that, on the face it, suggests significant improvements by the industry. But is this really the case? Sam Tomlinson from PwC discusses what conclusions we should draw, and what he believes should happen next.

When our 2020 study was published, the press headlines understandably focused on the unidentified, unattributable 15% (christened the “unknown delta”), although we were always clear that this was actually a secondary issue. In our view the primacy challenge was the widespread issues we encountered with data access and data quality, not least because addressing those was a prerequisite for any solution to the delta itself.

Our 2020 report therefore concluded that: (i) standardisation was urgently required across a range of contractual and technology areas, to facilitate the data sharing that is a key step towards more transparent supply chains; and (ii) all industry participants should collaborate to further investigate and reduce the unattributable unknown delta.

The resulting UK cross-industry Taskforce (comprising ISBA, AOP, IAB and IPA, plus individual members, with support from us at PwC) spent nearly two years developing a Toolkit – the Audit Permission Letter and Data Field List – designed to improve data access and data quality.

Our 2022 study for ISBA was a controlled test-and-learn of that Toolkit, conducted with a collaborative group of advertisers, agencies, tech vendors and publishers. As you might have already read, the headline results are certainly promising:

  • Data access: Improvements in data access successfully halved the time required to conduct the study to nine months (vs. 18 months for our 2020 study)
  • Data quality: Unlike 2020, in 2022 all tech vendors provided log-level data, and with substantially fewer missing data fields
  • Impression match rate: This marked improvement in data quality contributed to the proportion of ad impressions we could match from DSP to SSP rising to 58% (vs. 12% in 2020); and, moreover, these is good reason to believe the unmatched 42% would display similar results
  • Unattributable spend: The unknown delta was reduced to 3% (vs. 15% in 2020)
  • Publisher revenues: Rose by eight percentage points on a like-for-like basis.

Within our 2022 study, private marketplaces (PMPs) performed particularly well, with impression match rates above 70%, a delta below 1%, and publisher net revenues above 70%. These very encouraging results reflect positively on investments by advertisers, agencies, ad tech vendors and publishers in well-curated auditable PMPs.

So, does this mean the challenges in programmatic are solved? It’s not quite that simple…

First, this study is not representative of all programmatic. It comprised premium advertisers, agencies, tech vendors and publishers. We have no data on long-tail programmatic, but instinctively I would not expect it to look as clean as the results of our 2022 study. (It’s worth mentioning here that, in both 2022 and 2020, less than one-fifth of impressions were delivered to our study publishers, and in both studies the average number of websites being used was many thousands.)

Second, despite strong goodwill and collaboration on all sides, this 2022 study still took nearly twice as long as we hoped, due to the myriad complexities in legal permissioning and data transfers. This suggests more work is required before programmatic supply chain audits can become commonplace.

Third, even at this premium end of programmatic, an iterative approach was required in which our PwC team (of data scientists, ad tech experts, and qualified auditors) worked closely with the tech vendors to identify data discrepancies and re-run reports. It was this hard work, on all sides, that helped deliver such promising results.

In summary:

  • Yes, the premium end of programmatic has undoubtedly made real operational improvements, which is clearly good news;
  • But, no, these results should not be assumed to be representative of all programmatic; and
  • More work is still to be done.

Specifically, we see on ongoing role for the industry to continue its collaboration via the Taskforce: refined protocols are required for data access, data quality, data retention, and data transfer, if audits are to become routine.

In the meantime, our conclusion is that this study shows that, with effort, premium programmatic can function really well.  These results show what the industry should aspire to, rather than what is generally happening.

Finally, I’d like to express our gratitude to everyone at ISBA and the Taskforce, and to all the participating advertisers, agencies, tech vendors and publishers; and also to my own PwC team who delivered this work. There’s undoubtedly still plenty of work to be done, but it’s really heartening to see the progress since 2020.