Programmatic advertising in 2026: UK platforms and agencies

What to take away

The Digital Markets, Competition and Consumers Act 2024, the pace of Google Ads releases, the shift of video buying toward automated outcome models and the CAP Code compound. The Act created a conduct regime for firms with strategic market status. Google Ads kept shipping interface and feature changes at pace. Video buying moved further toward automated outcome models.

And the ad codes stayed the fixed point that all of it must satisfy.

That combination changes what a programmatic agency is paid to do. Media buying still matters, but the defensible work now sits in governance, measurement design and the ability to explain a decision to a client, an auditor or a regulator.

This guide treats the year ahead as a set of decisions, not a single number. Forecasts age quickly; triggers, owners and definitions do not.

Before a planning cycle starts, know which way demand and pricing are moving. Our programmatic advertising market outlook for the UK covers the signals to track when a client asks whether a budget is realistic.

Regulation and platform conduct

Regulation decides what platforms may do and how their conduct is policed. Two threads matter for buyers: the conduct regime created by the 2024 Act, and the steady flow of platform releases that alter how campaigns run.

What the DMCC Act changes for buyers

The Digital Markets, Competition and Consumers Act 2024 received Royal Assent in May 2024. It gives the competition authority powers to designate firms with strategic market status and to impose conduct requirements on them.

The regulator is the Competition and Markets Authority, which in January 2025 began the first strategic market status investigation under the regime, into Google's general search and search advertising.

For buyers, the practical effect is transparency. A designated platform can be required to explain how it ranks, prices and shares data. That matters when a client asks why a cost per acquisition moved and the answer sits inside a black box.

Do not read the Act as a price control. It is a conduct regime. The near-term work is record keeping: know which platforms you depend on, what share of budget each takes, and what you would do if terms changed.

Build that record as a dependency register. For each platform, note the contract term, the notice period, the share of client spend it carries and the cost of moving elsewhere. Review it twice a year.

Concentration is the number that surprises clients most. If one platform carries most of a brand's spend, that is a commercial risk as much as a media one.

Keep the register where the team can find it, not in one person's inbox. When a client's procurement team asks how the agency manages platform risk, that document answers the question.

The Act also modernised consumer protection rules, including measures on unfair commercial practices and hidden fees. Price claims in programmatic creative are therefore a consumer law question, not only a code question.

That matters at scale. One price claim can run across thousands of placements, so the offer behind the ad deserves more review time than the banner itself.

Platform change as a standing item

Google Ads is the clearest example of continuous change. Features appear, migrate and retire, and the announcements page is the primary record. If nobody owns that page, the first signal is usually a broken report.

Give one person the job of reading release notes monthly and mapping each change to a live campaign or a client report. That is cheap. Discovering a bidding change during a quarterly review is not.

Keep a short log: what changed, which accounts it touches, who is checking it and by when. A log turns a public announcements page into an internal audit trail.

Not every release needs action. Sorting them into ignore, watch and act takes minutes and stops a team reacting to changes that never touch its accounts.

Other platforms publish release notes on similar cycles, and the same ritual applies. Pick one owner and one document rather than spreading the job across the team.

This is also where regulatory and platform risk meet. A tool change can alter how a price is shown or how a claim is targeted. The CAP Code applies to the output either way, so compliance review belongs in the same cycle as platform review.

The platforms and agencies behind a UK plan

Almost every UK programmatic plan is bought through a demand-side platform. Display & Video 360 is Google's enterprise buying platform, The Trade Desk is the largest independent DSP, and Amazon DSP opens Amazon's own retail inventory to the open market. Microsoft Advertising's Xandr and Criteo are the other names a UK buyer meets, the first on Microsoft inventory and the second across retail media.

On the sell side, publishers route inventory through supply-side platforms and exchanges. Magnite, PubMatic, Index Exchange and OpenX appear most often in UK supply paths, while Google Ad Manager remains the ad server behind a large share of UK publisher inventory. Ask for the supply path, and the fee taken at each hop, before anything is signed.

Agencies sit in the middle. The largest UK advertisers buy through a holding group: WPP's GroupM, Publicis Media, Omnicom Media Group, Dentsu and Havas Media each own major UK agencies and run their own programmatic trading and analytics teams. Independents and specialist programmatic agencies buy on the same platforms, so the choice of agency usually changes the trading terms rather than the technology.

Verification sits on top of delivery. DoubleVerify and Integral Ad Science are the vendors most UK buyers use to check viewability, invalid traffic and brand safety, and both publish the methods behind their scores. Agree before launch whose numbers settle a dispute.

Trade bodies set the shared standard. IAB UK publishes the guidance the market works to, ISBA represents advertisers and the IPA represents agencies. On data, programmatic targeting in the UK sits under UK GDPR and the Privacy and Electronic Communications Regulations, with the Information Commissioner's Office as the regulator; the ICO publishes guidance on consent, cookies and identifiers.

Formats, inventory and measurement

Format change is slower than platform change and usually more consequential. A new placement changes who sees the ad, what they expect from it and which metric proves it worked.

Video and the move to outcome models

The IAB's August 2026 paper on AI-powered video outcomes describes a shift from buying impressions toward buying defined results. In practice the model is asked to find the outcome, and the buyer's job becomes defining the outcome well.

That is harder than it sounds. A vague objective produces a vague optimisation. Write down what counts as a result, over what window, and what is excluded, before the campaign goes live.

Put that definition in the campaign brief, not in a slide shown once. The people trafficking the campaign need to see the same words the model is optimising against.

Video is also where brand and response budgets collide. Keep the two objectives separate in reporting, or the model will optimise toward whichever signal is easiest to find.

Measurement windows deserve the same care. A short window flatters response campaigns and penalises anything that works slowly, so agree the window with the client before results arrive.

Agree who owns the measurement source before the campaign starts. Two teams reporting the same campaign from different dashboards is a common cause of client doubt.

Supply deserves attention too. Ask where the inventory comes from, which placements are excluded, and whether the creative holds attention at the length actually served.

Formats outside the feed

Not every placement is a banner in a feed. Sponsored Messaging delivers an ad into a LinkedIn inbox, which behaves differently: fewer impressions, higher attention per impression, and a different compliance surface because it reads like correspondence.

Use it where the audience is small and known, not where you need reach. Measure it against a reply or a qualified conversation, not a click-through rate benchmark borrowed from display.

Direct formats also pull copywriters into the compliance conversation. Someone who signs off client correspondence should see the ad before it ships, because a claim that reads well on a banner can read differently in an inbox.

More broadly, format choice should follow the decision the buyer is trying to influence. A format that cannot be measured against that decision is a cost, not a channel.

Programmatic delivery also multiplies creative risk. A single asset can appear in thousands of contexts, so the review needs to consider the worst placement, not the average one.

When a format has no benchmark, run a bounded test. Fix the budget, fix the end date, and agree in advance what result would justify another round.

Skills, scenarios and decisions

Planning assumptions are worth little if nobody owns the response. This section covers the people question and the rehearsal that turns a risk list into an action.

The skills gap is a planning input

Automation moves the human task from operating a console to defining rules, auditing outputs and explaining trade-offs. That is a different skill set, and it is scarcer. Our programmatic advertising skills forecast looks at which roles are growing and which are being absorbed.

Plan headcount around governance and analysis, not around manual optimisation. A team that can read a log, challenge a model and write a clear client note is worth more than one that can only move budget between line items.

Training is usually quicker than hiring. A trader who already knows the accounts can learn objective design and evidence writing in a few months, and that knowledge stays with the agency.

Write the job description for the work that will exist in 2027, not the work that existed in 2023. Otherwise the team fills up with skills the platform now performs itself.

Blended roles are emerging, where the same person writes the brief, sets the objective and reads the result. That structure suits small teams with a handful of large accounts.

Scenarios worth rehearsing

Write down three scenarios now. A major platform changes terms or fees. A regulatory intervention forces a reporting change. An automated buying model underperforms and the client asks for evidence.

For each, decide who acts, what the client hears and how quickly. Our programmatic advertising risk scenarios for the UK set out how teams can rehearse these before they happen.

Scenario planning is not pessimism. It is the difference between a considered response and a panic meeting.

Give each trigger a number where possible. A rule such as 'review if cost per acquisition rises by a fifth across two consecutive months' is testable, while 'review if performance drops' is not.

Rehearse once a quarter with the people who would actually be in the room. A plan nobody has read is a document, not a capability.

Decide in advance who calls the client and what the first version of the message says. Speed of explanation usually matters more than completeness on day one.

A decision table for the next four quarters

Use the table by finding the row that matches the situation in front of you, then checking the response and the trap beside it.

What to watch and when to act

Set review triggers rather than calendar dates. A trigger is a specific event: a designation decision, a platform deprecation notice, a sustained cost shift beyond an agreed band.

For the 2027 planning cycle, our UK programmatic advertising trends guide sets out what to build into next year's plan and which assumptions are worth stress-testing with a client.

Review the trigger list each quarter so stale items do not sit there unused.

When a trigger fires, the response should already exist. That is the whole point of writing it down in advance.

How to run the next planning cycle

The operating model matters as much as the forecast. Delivery, trafficking, QA and reporting all need owners, and automation changes what each owner does rather than removing the role.

For each active campaign, write the objective definition: the outcome, the window, the exclusions and the reporting source. This is the document that makes an automated buying decision auditable, and our programmatic advertising operations and delivery guide shows how it feeds trafficking, QA and reporting.

Next, set the compliance path. Every creative and every claim should have a named reviewer and a record of the decision. Keep that record even when the answer is yes.

Finally, agree the client narrative. Clients do not need every platform detail, but they do need to know what could change, what you would do, and what it would cost. A short quarterly note covering those three points prevents most difficult conversations.

Sequence the work so the objective definitions and the compliance path are finished before any creative is briefed.

Put a date in the diary to review the whole plan, not just the media results. Quarterly is enough for most accounts.

Budget for the work. A team paying £400 a month for a reporting tool, for example, still needs analyst time to interpret it.

Common questions

Is programmatic advertising still growing in the UK?

The direction of travel is continued growth, but the mix is shifting toward video, retail media and automated outcome buying. Treat any single growth figure with care unless it names its source, sample and date.

What does the DMCC Act mean for a small agency?

Directly, little, because the conduct regime targets firms with strategic market status. Indirectly, more transparency from the largest platforms, which should make fee and ranking questions easier to answer.

How often should we review platform changes?

Monthly is a sensible default for reading release notes, with an immediate review when a change touches bidding, targeting or reporting on a live account.

Do automated video buying tools remove the need for a media planner?

No. They change the task from manual optimisation to objective design, auditing and interpretation. The planner's value moves to defining the outcome and challenging the model's output.