What programmatic display is, and why it needs scrutiny
Programmatic display is the automated buying of ad space across websites and apps, usually auction by auction, through a chain of technology platforms. Instead of booking a banner on a named site, an advertiser describes who it wants to reach and the systems find those people wherever ad space is available.
It is a large and growing market. IAB UK's Digital Adspend study put UK digital advertising at £40.5 billion in 2025, up 10% year on year, with video at £9.3 billion and growing 20%; the first half of 2026 reached £21.2 billion, up 13% (IAB UK, 2026). Display and online video account for a substantial share of that, and much of it is traded programmatically.
The same automation that gives reach also makes it hard to see where money goes. The ISBA and PwC study of the UK programmatic supply chain, published in 2020, found that publishers received only 51% of advertiser spend, and that around a third of supply chain costs could not be attributed at all, a gap the study called the unknown delta (The Drum, 2020). The study covered about £0.1 billion of spend across 15 advertisers, and later work reported improvements, but it remains the foundational UK evidence on the problem.
The waste problem, in figures
US research has found similar issues. The Association of National Advertisers' programmatic transparency study, run with Kroll and updated in December 2023, found that at most 36 cents of every dollar spent reached the publisher (Legal 500, 2023). The average campaign ran on over 40,000 websites, and made-for-advertising sites, built mainly to sell ad space, took 21% of impressions and 15% of spend. The sample was 21 marketers and 12 supply chain companies, all volunteers, so it is not a census of the market.
There is progress where advertisers act on the evidence. The ANA's first quarterly benchmark in 2024 found that spend on made-for-advertising sites had fallen from 15% to 4%, and that the average campaign now ran on about 23,000 sites (AdExchanger, 2024). That benchmark analysed log-level data for 11 brands, so again it is a small sample, but the direction is clear: buyers who inspect their placements and cut the long tail waste less.
Whether an ad can be seen at all is another issue. The Media Rating Council's viewability standard counts a display ad as viewable when at least 50% of its pixels are on screen for one continuous second, or two seconds for video (Publift, 2026). Integral Ad Science's latest Media Quality Report found global display viewability of 67.9%, against 79.7% for video, which means around a third of display impressions still fall short of that low bar (Advanced Television, 2026).
Define the audience by behaviour
Start with the people who matter most, then describe them by what they do rather than only who they are:
- Visitors who viewed key pages, such as pricing, services or product pages.
- Lookalike audiences built from your best customers or most valuable accounts.
- In-market signals that show a buyer is actively comparing options.
- Contextual targeting, where your message appears beside relevant content.
- Named-account lists for B2B, matched to the companies you most want to win.
How audience buying works, and the data behind it
Each audience type does a different job. Site visitors and customer lists are small but valuable, because these people already know you; they suit retargeting and offers. Lookalike and in-market audiences are larger and colder; they suit awareness and consideration, where the aim is to introduce the brand to people who resemble your best customers. Contextual targeting places ads beside relevant articles or videos, which needs no personal data and often reaches people at a moment when the topic is already on their mind.
The audience definition should come from the business, not the platform's default segments. Start with what you know about your best customers: the sectors they work in, the problems that bring them to you, the pages they read before enquiring and how long they take to decide. Then choose the targeting methods that best match those facts, and test them against each other rather than running them all at once.
Audience buying depends on data, and the rules around that data keep moving. In April 2025 Google confirmed it would keep its current approach to third-party cookies in Chrome and would not introduce a separate prompt asking users to choose, after receiving divergent views from publishers, regulators and the ads industry (Google Privacy Sandbox, 2025). Chrome's Incognito mode continues to block third-party cookies by default, and other browsers already restrict them.
The practical lesson is not to build a plan that depends on any single tracking method. First-party data, such as customer lists and site visitors who have consented, contextual targeting that needs no personal data, and careful use of platform audiences together give a more resilient base than third-party cookies alone.
Set frequency, creative and placement controls
Awareness needs reach and a frequency cap, so the message is seen but not repeated to exhaustion. Consideration needs proof, such as a case study or a benchmark. Retargeting needs a specific reason to return, which is where retargeting adds most value.
Set brand safety rules, exclude sites that do not suit the brand, and review placement reports each month. Quality matters more than volume: a smaller number of well-chosen placements usually outperforms a broad network that is hard to inspect. Placement still matters inside an audience buy, because reaching the right person on a low-quality site full of ads does little for the brand; audience targeting and placement controls work together rather than as alternatives. Ask for a list of the domains and apps where your ads ran, and use inclusion lists or curated private marketplaces where you can.
For example, a professional services firm might start with an inclusion list of a few hundred business and trade titles, a cap of a few impressions per person per week, and three creative versions matched to awareness, consideration and retargeting.
Measure beyond clicks
Click-through rates undervalue display, because most of its effect comes from people who see an ad and act later. Use a holdout group, which is a comparable audience with no ads, to measure the lift in branded search, site visits and sales. Google's Conversion Lift tool works this way, comparing conversions from people shown ads with a group held back, and reporting incremental conversions and incremental return on ad spend (Google Ads Help, 2026). It also warns that relative lift can look inflated when the control group has few conversions, so read the absolute numbers too.
Marketing science methods make that comparison more reliable, and they show where the next pound should go. Useful measures include:
- Incremental conversions and cost per incremental conversion from a holdout test.
- Viewable impressions and the share of spend on known, quality domains.
- Lift in branded search and direct traffic in exposed areas or audiences.
- For B2B, engagement from target accounts and their progress through the pipeline.
Keep display in proportion, and what to ask a supplier
Display works best alongside search and social, not as a replacement for them. Its job is to make the brand familiar, so that search and sales conversations are easier when they happen. Our display service plans it within the wider media mix.
When you choose a supplier, ask what share of your spend reaches publishers, whether you will see log-level or domain-level placement data, how they exclude made-for-advertising sites, what viewability they target, and how they will prove incremental impact with a holdout rather than relying on platform-attributed conversions.
Put this into practice
Book a discovery call and we will review your channels and results first, then agree what to test.
Plan your displayQuestions and answers
Is display worth it for B2B?
Often, for reaching named accounts and roles. Measure it against the accounts and enquiries it helps, not only the clicks it produces, and use a holdout group to see whether it changes behaviour.
What is a holdout test?
A holdout test keeps a comparable group without ads, so you can compare results and see the lift the advertising creates. Google's Conversion Lift tool is one example, reporting incremental conversions rather than attributed ones.
What are made-for-advertising sites?
They are sites built mainly to sell ad space, often with heavy ad loads and little original content. An ANA study found they took 21% of programmatic impressions in its sample, so excluding them is one of the quickest ways to cut waste.
What counts as a viewable display ad?
Under the Media Rating Council standard, a display ad is viewable when at least half of its pixels are on screen for one continuous second. Video needs two seconds. It is a minimum bar, not proof that anyone noticed the ad.
Will the end of third-party cookies stop programmatic targeting?
Google confirmed in 2025 that it would keep its current approach to third-party cookies in Chrome. Even so, other browsers restrict them, so plans that combine first-party data and contextual targeting are more resilient.
References
- IAB UK, "IAB Digital Adspend", 2026
- The Drum, "Big hole in the value chain: one third of ad tech costs unattributable, finds Isba", 2020
- Legal 500, "ANA study finds over $20 billion in 'efficiency gains' possible in programmatic advertising spend", 2023
- AdExchanger, "The ANA says advertisers are spending way less on MFA, but programmatic ain't fully transparent yet", 2024
- Publift, "MRC viewability", 2026
- Advanced Television, "Report: Video viewability outperforms display" (IAS Media Quality Report), 2026
- Google Privacy Sandbox, "Next steps for Privacy Sandbox and tracking protections in Chrome", 2025
- Google Ads Help, "Understand your Conversion Lift based on users measurement data", 2026

