In this exclusive article for ExchangeWire, Jonas Jaanimagi (pictured below), technology lead at IAB Australia responds to ISBA’s recent study on the programmatic supply chain, and contemplates how the industry should respond to its findings.
Last week in the UK the results of the Programmatic Supply Chain Transparency Study were revealed, put together by the ISBA (Incorporated Society of British Advertisers) and PwC over a period of some 18 months. The intention of the report was to try and respond to the ISBA members’ question – what does my programmatic supply chain look like and how can I assess its value in terms of working media?
At a glance the results are not pretty and the resulting headlines in global industry media have inevitably painted a picture of rank inefficiencies, rampant malpractice and increasing levels of mistrust. However, the devil is always in the detail and our honest response, upon some reflection, is that picking away at the methodologies of this study is unhelpful and it’s better instead to constructively leverage reports such as these as invaluable awareness drivers for the industry to help itself instigate genuine change. With this in mind, we’ll try and constructively dive a bit deeper in terms of the related issues, provide some general recommendations and then highlight what we intend to propose to our members and industry partners as practical next steps here in Australia.
Before we look at the results, we need to focus on some definitions and discuss value vs. cost. For instance, ‘working media’ is a term traditionally defined as the percentage of an advertisers’ spend that successfully delivers their messages to their intended audiences. Historically, the higher end of non-working media expenses for services such as production and agency fees in advertising, was always around 20% – but to purely classify the monies received by publisher as ‘working media’ as a standard convention is incorrect and remains a huge misunderstanding of a key difference between direct sales based upon signed contracts and real-time bidding on open market inventory.
Much of the value of ad tech isn’t simply delivery and reporting (e.g. like a generic ad-server) but in allowing buyers to identify specific audiences, appropriate environments and data-driven opportunities in real time from a very wide range of supply sources. The commercial and attributable value-add of these dynamic targeting attributes are almost always neglected in these studies and treated solely as costs, ignoring the benefits. Cutting-edge advertising capabilities have revolutionised advertising for buyers and are not always enabled by the publishers whose content intended users happen to be consuming at the time that the ad is served. Yet the calculation of ‘working media’ treats these clear benefits solely as costs and incorporates only the publisher revenues received from these transactions as the headline value, thus ironically rendering the total stated and terminology redundant. If buyers are utilising optional tools or services that they aren’t aware of, don’t have contractual oversight on and/or simply don’t understand the incremental value they add – then retrospectively blaming these value-add capabilities for any of the costs incurred voluntarily still remains a serious issue.
As for the headline results from the study themselves:
- 51% of advertising spend (with an overall range of between 49% and 67%) was being received by publishers – and classified as ‘working media’
- There was a 15% unknown delta of unattributable costs within the supply chain
- From 267m impressions served, only 31m (12%) could be successfully matched
As promised, we’ll now dive into the details of the key figures as per the above and highlight some key points, proposed commitments and recommended next steps. We’ll try and stick largely to the operational and avoid anything overly contractual, as the commercial agreements between entities is up to them – however we regularly try to provide best practices in this area and always recommend regular review and close scrutiny over any vendor contracts.
Retrospectively auditing log-level data and trying to match without utilising the latest transparency standards was always going to be very messy
It’s a natural and logical approach, wherever money is involved, to try and run an auditing process. To objectively match any monies spent by buyers directly against the monies received by sellers and highlight the ancillary costs of any intermediaries used to enable, deliver, measure or enrich these transactions. Digital advertising has evolved so organically and rapidly over the past 20 years or so that anyone involved at the coal-face operationally in our industry will have grimaced trying to imagine auditors objectively approaching this project with anything resembling standard auditing practices. That’s not to criticise the effort involved (although the details of the methodologies are clear from the public documents) or to blindly defend the programmatic infrastructure, but this issue is hardly a new one – particularly since the ubiquitous adoption of programmatic globally and the explosion of header bidding on desktop, and increasingly mobile and video.
Every programmatic seller and buyer has faced versions of this in their internal operational and financial practices and the ability to interpret the nuances between the various siloed sets of data is why technical and product expertise and experience are so vital in digital advertising. It often takes a long time and a lot of effort to review the legal details, manually interpret the log-level data files, understand the nuances of different platforms, try to automate as many of the collection processes as you can, establish logical rules and then attempt to reconcile cleanly. Trying to do this objectively and retrospectively across all the moving parts, with thousands of supply sources and dive into the various contracts was always going to be a nightmare. As a result, the 12% match-rate, whilst very low, is not a massive surprise and the time taken wading through the contracts, interrogating data and costs incurred as a result should have been better anticipated by the auditors. The log-level data permissioning and access procedures on their own, without any prior notice or precedence, will have been back-breaking. Programmatic is not a simple and defined product or channel, it’s an inevitable mechanism of automation with an infrastructure underpinned by ever-evolving technical innovation and as a result of its organic evolution needs to be objectively overseen by the most up-to-date standards that have been mutually agreed by the industry. Otherwise and inevitably complexity results and due to the pace of innovation often only worsens.
The IAB Tech Lab is the non-profit entity that globally oversees the protocols of programmatic (known as OpenRTB) which were first established almost 10 years ago. The most recent version of these protocols is OpenRTB 3.0 which was released in November 2018 – and the core focus of these protocols has been almost entirely on enabling transparency through the supply chain, allowing the integration of billing reconciliations and reducing fraud. All have been critical issues for the industry for some time and are inevitably highlighted as crystal clear call-outs from this study. Unfortunately however, having had the product and technical specialists in our industry all mutually agree that v3.0 protocols were the best way forwards for programmatic trading, the wider industry has since shown little interest in investing the engineering time and effort required for the upgrading and full adoption. Some other accompanying recent standards are at least mentioned in the report (sellers.json and OpenRTB SupplyChain object) so one has to assume that whilst there is some knowledge of them they were not insisted upon prior to the auditing process by the consultants involved. These, along with a publisher standard called ads.txt, which they build upon, are now critical tools in verifying and identifying programmatic sellers and re-sellers – and ultimately in being able to map the path of supply. We’ll dive a bit deeper into these in a moment…
Without even an attempt at enabling full clarity over the data prior to the reconciliation efforts there were always going to be some serious gaps, however the 15% of costs being unexplained remains a real concern. Our initial assumptions on these are below, and we hope they are useful. We’ll be working more closely with some of our specialist councils here in Australia to see if we can collaboratively provide even more clarity on this.
- Platform reporting discrepancies persist – results in a simple disparity between what a DSP charges and an SSP pays.
- Non-transparent vendor (i.e. DSP and SSP) rebates or fees – for example there may be partner fees that could not be factored-in.
- Resellers within the chain that could not be identified – without the latest standards being fully utilised the involvement of resellers may not have been clearly identifiable within the process.
- Post-auction bid shading – this practice can make it hard to clearly identify the final amount of a winning bid.
- Post-auction financing arrangements – some trading deals may not have been initially factored-in and many are triggered by volumes and/or over different time periods.
- Limitations in data sets resulted in estimations – as so much data could not be matched the error rate on the estimations could be significant. Without uniformly adopting the IAB Tech Lab’s OpenData standard as a lingua franca between all the different parties this process will have been made even harder.
- Foreign exchange fluctuations – RTB is predominantly transacted in USD. The starting and ending currency in usage here is GBP and the currency conversions would occur more than once and the conversion methodologies vary.
- Ad fraud – remains a serious issue globally and without more insights into the setup one cannot discount the possibility of some of the bids being siphoned away by criminals, particularly as thousands of supply sources were utilised. Utilising the very latest standards and working closely with technology vendors on this topic is absolutely critical for both buyers and sellers, regardless of any perception of the inventory being labelled as premium.
As many of the required standards already exist let’s work on the industry now accepting the critical importance of these standardisations, enable adoption at-scale and commit to expediting the release of any related standards still in development
The importance of the OpenRTB protocols and its layered specification approach (OpenMedia) cannot be underestimated and we’ve recently worked hard here in Australia to drive awareness around as to why. Essentially a key raison d’être of non-profit industry bodies such as all the global IABs and IAB Tech Lab is to provide, support and enable full transparency and trust after all.
However, as mentioned there has been a collective failing here and their adoption requires more than simply deep product understanding and technical ability, as there needs to also be an accompanying commercial awareness of why they are important. OpenRTB 3.0 has been available for 18 months now and whilst their importance has been recognised by the technical specialists involved in their collaborative evolution it’s perhaps only now that we can see the corresponding commercial value of the related technological transparency and trust. The good news is that the majority of the initial work has already been done, however there now has to be a commitment to all of the businesses involved in the supply chain to the engineering time and effort required to upgrade. For us to continue to delay and devalue these critical standards as an industry would be irresponsible.
As well as OpenRTB 3.0 there are other examples of critical standards that are currently available. SupplyChain Object, ads.txt and sellers.json are standards which enable buyers to see all the parties involved and that were paid as part of an impression opportunity from beginning to end – and ultimately can act as a record for every impression. OpenData is a standardised and consistent language for vendors, advertisers and publishers to utilise in all campaign reporting. This will provide publishers, agencies, and data management vendors with a common language and mapping tool intended to improve workflow in day-to-day campaign analytics processes.
There are also some other examples of standards in development, such as ads.cert – which enables signed bid requests to confirm the valid ownership or source of inventory in real-time and verify related critical data points such as domain, IP, format, etc. There is even currently a v1.0 beta of this standard already publicly available, but it requires the OpenRTB 3.0 upgrade. The IAB Tech Lab are also working on a buyers version of sellers.json so as to enable similar levels of transparency on the buy-side, as currently exist on the sell-side – and potentially make any retrospective audit process much easier.
However, to fully include something resembling a transaction ID into the programmatic protocols and allow for improved retrospective auditing remains a chicken and egg scenario. Without there being a full commitment now by the industry to adopt the suite of programmatic standards that currently exist, the confidence in non-profit industry bodies being able to evolve what we have and invest in even better iterations and standards is simply impossible. We’re all in this together and as adoption is not mandatory it has to be a collaborative industry-wide commitment on the back of studies such as this – and the resulting commitment has to be total, not only supported by certain specialist divisions within our industry.
Australia can lead the way in a global pilot with fully transparent standards in-play and executed via a pre-determined and mutually agreed set of recommended best practices.
IAB Australia has been working very closely with the IAB Tech Lab for some time now, but more importantly Australia has also proven that its industry bodies can co-operate well on joint projects. We are in the process of collaborating with the MFA (Media Federation of Australia) and AANA (Australian Association of National Advertisers) to update The Australian Digital Advertising Practices first released 18 months ago to its latest version. These have been created specifically for advertisers in Australia to facilitate better relationships with publishers, agencies, platforms and technology vendors, and to foster shared responsibilities for digital ad spends and outcomes. We have also regularly produced educational documents aimed at building education and awareness in programmatic, such as the Programmatic handbook, Auction Mechanics playbook and AdTech buyers guide. These have been collaboratively created by representative councils such as the Executive Technology Council, Standards & Guidelines and Ad Effectiveness which contain a very healthy mix of publishers, technology vendors, advertisers and agencies as members and contributors.
As a result we feel very confident that we could work together to enable a local pilot which attempts to answer the same question as the ISBA study, but with a different approach (most likely leveraging some communal goodwill in terms of supply and demand!) and to include as many of the IAB Tech Lab standards in its setup as possible. We have already formed a related working group to at least review the potential of this and are communicating regularly with the other relevant local bodies.
Regardless, we would immediately recommend the below as a result of this study – and are happy to provide more information on these points as and when required.
- Widespread immediate adoption of sellers.json (identifies direct sellers or intermediaries) and SupplyChain Object (reveals hops between sellers and buyers)
- A commitment of adoption of OpenRTB 3.0 by December 2020
- Full adoption of OpenData and the industry-wide standardisation of mutually agreed permissions access to data between publishers, SSPs, DSPs and agencies
- Widespread adoption of the IAB Tech Lab Taxonomies (for audiences, content and products) to enable consistency and assist with retrospective reporting
- DSPs should insist upon using separate trading seats for each of their clients to improve reporting
- Consistent usage of Supply Path Optimisation and alignment on best practices, so as to minimise bid duplications and improve the overall efficiency of RTB
- Review the inclusion and adoption of some form of transaction ID within the OpenRTB protocols to enable cleaner retrospective audits
- Expedite the final development and release of IAB Tech Lab’s proposed buy-side transparency standards
- Full testing of ads.cert once OpenRTB 3.0 has been more widely adopted
- Contractual best practices in relation to data access, to build upon the recommendations within our AdTech Buyers Guide.
Another critical project underway also being led by IAB Tech Lab is project rearc which is a global initiative to rearchitect digital advertising with the impending demise of third-party cookies over the next 18 months or so. This will ultimately act as a form of a reset on much of the supply chain plumbing, as data plays such a key role in programmatic transactions. Therefore the timing is also an opportunity for reflection and review for us all as well for the expedition and or adoption of critical standards. IAB Australia’s direct involvement in the technical working groups of this global collaborative effort also helps provide our members with direct insights into the potential developments in how the industry intends to rearchitect itself from a technical perspective and will enable us to also ensure that the issues raised by this ISBA/PwC study are fully considered.
We look forward to further engagements with the industry on these types of constructive insights as we work with IAB Tech Lab, our members and the industry-at-large to try and ensure that programmatic doesn’t continue to ‘underlive it’s life’ as a critical mechanism for advertising moving forwards.
For more information on these IAB Tech Lab standards from the IAB Australia website please click here