By Raman Sidhu, Co-Founder and Chief Revenue Officer at Beemray, now part of Verve Group
As the hunt to replace tracking cookies gathers pace, advertisers must now identify and experiment with the alternatives.
And over the next 18 months, at the end of which third-party cookies are expected to be fully deprecated from mainstream use, adland will have plenty to road test – including cohorts, clean rooms, universal IDs and a host of new privacy enhancing technologies.
Each has their benefits and pitfalls, and some are more developed than others – but at this stage there is one solution that has the attention of the early-adopters who want to hit the road running: contextual programmatic advertising.
But like any new technology, it is wise for those thinking about investing in its use to know what questions to ask.
Not all tech is equal
The advertisers who have been experimenting with contextual targeting have already discovered it offers both performance and scale within a brand safe model, and can also be used as a proxy for audiences.
As a result, we’ve seen it perform better than cookie-based solutions across KPIs for brand and performance advertisers.
That’s lots of boxes ticked – but where advertisers will need to begin digging deeper is linked to the sophistication and efficiency of the underlying tech.
In the instance of the former, it will be crucial advertisers and agencies ask whether the technology that underpins contextual programmatic merely takes a ‘keyword approach’ – which works, but I would argue is too simplistic – or whether the tech is tuned well enough to understand the multi-dimensional aspects of a webpage.
Advertisers, therefore, should check whether the tech also understands semantics, sentiment and emotion – the same crucial factors a human would check before deciding whether an ad placement was a wise choice.
As contextual targeting continues to gain momentum in a privacy-first world, it’s important for brands and agencies to choose technology that goes beyond off-the-shelf offerings, and instead focus on deeper real-time contextual analysis.
Test for efficiencies
There are three steps involved in programmatic contextual targeting: the discovery of new content; analysis to extract the contextual value; and activation, which makes that context available for targeting within the DSP.
However, advertisers should be mindful to check the reality of latency within this process. Our own analysis shows, on average, the industry waits up to 24 hours to activate new data, losing 70% of the traffic which is most valuable for advertisers and publishers.
Supply chains can be burdened by quite severe data inefficiencies, consequently we have seen media investment suppressed against relevant, high quality and newly published content.
The reason behind this is most likely the absence of a real-time and end-to-end data flow between discovery, analysis and activation. In simple terms, this means advertisers and publishers risk missing out because content cannot be matched and refreshed quickly enough.
Therefore, I would urge advertisers to get under the hood and benchmark latency as part of their due diligence alongside their questions about the sophistication of the machine learning that sits on top of the core tech.
Customization
Advertisers should also be mindful of which framework they are plugging in to, and ask how much customization they require.
Brands that need real-time responses to inform planning, creative and buying decisions will find the best performance and efficiencies in a model that is plugged directly into the programmatic supply chain – rather than having a data management platform build an audience which is then taken to a DSP.
This is the opposite of what I’d describe as a cookie-cutter approach, and it’s not right for every business. But for brands who want to learn and analyse in-the-moment and respond back to the most relevant live bids, they should be clear about what off-the-shelf adtech can achieve compared with something more boutique.
Likewise, if an advertiser requires a specific segment, then it should really be specific to them – and not just be translated into the closest off-the-shelf template. Advertisers should be clear about their own needs here, and seek a solution that fits them best.
Evolution of contextual
Programmatic contextual advertising is changing as the technology is fine-tuned, and as more advertisers use it and supply feedback.
This means we’re looking at ways to overcome challenges such as managing frequency without the use of personal identifiers, which could see the introduction of data modelling techniques.
But the real evolution will be delivering audience and creative insights from contextual data – and using contextual signals to personalize advertising messaging.
To achieve all of this, and more, we will need to take a unified view of the challenges and solutions. And it will also be essential to maintain an open dialogue with every part of the supply chain, ensuring what we are building works for the most demanding and innovative advertisers.
Collectively, adland is in the process of rebuilding the commercial internet for a post-cookie world. Let’s all ensure we push for the best possible outcomes early on.
