A 90-day programmatic advertising plan for a known brand or a test budget

What to take away

What a 90-day plan is for

A 90-day plan suits a client who wants proof before a year-long commitment. It also suits an agency testing a new channel, a new region or a new format.

The plan needs one owner, one budget and one decision date. Without those three, a quarter becomes an open-ended pilot.

Keep the scope narrow. One product, one audience and one conversion event give a clean read. Adding a second market in week six usually destroys the comparison.

The programmatic advertising strategy and planning guide sets the wider objectives this quarter should sit under.

Two starting points: a known brand or a test budget

A known brand and a challenger testing a budget run the same calendar but not the same quarter. The known brand usually has an ad server, a consent stack and a measurement partner already in place, so the quarter goes on incrementality, brand safety and brand lift rather than on proving that programmatic works at all.

For a known brand, name the verification vendor in the brief — Integral Ad Science or DoubleVerify — and commission a brand-lift study from Nielsen or Kantar. Keep a control group that sees no programmatic creative, so the result is incremental lift rather than last-click credit.

For a team running a test budget, do the opposite. Keep the stack small, buy open-marketplace inventory, and prove one cost per acquisition before paying for managed service, private deals or a second platform.

The platforms and partners the plan should name

Demand-side platforms do the buying. In the UK market the usual choices are Google's Display & Video 360, The Trade Desk, Amazon DSP and Yahoo DSP. The platform you already license normally wins, because migrating mid-quarter costs a month of learning.

Inventory arrives through supply-side platforms and exchanges such as Google Ad Manager, Magnite, PubMatic and Index Exchange. Private marketplace and programmatic guaranteed deals are negotiated with those SSPs or directly with publishers, and each one needs a deal ID before the campaign can be built.

The ad server sits between the DSP and the reporting. Campaign Manager 360 is the common choice, with a clean room such as Google Ads Data Hub used to match first-party data without moving it.

Agencies and trading desks buy this inventory on a client's behalf. The largest UK trading desks sit inside GroupM, Publicis Media, Omnicom Media Group and Havas Media, while independent desks and specialist programmatic agencies compete on transparency and on access to inventory.

Weeks 1 to 4: set the rules

Write the audience profile first. State the signal you will buy on, such as a site visit, a CRM match or a job title, and the exclusions.

Then write the brief on one page: objective, product, market, platforms, deal IDs, budget ceiling, primary metric and guards. Week nine is decided by what this page said in week one.

Then check the data rules. The ICO's guide to individual rights explains access, erasure and objection rights, which shape how long you can hold audience segments. The legal anchors are the UK GDPR, the Data Protection Act 2018 and, for cookies and similar storage, the Privacy and Electronic Communications Regulations (PECR). Consent has to be collected before a non-essential tag fires, so the consent management platform must be live before the campaign build starts.

Advertising standards apply too. The ASA and CAP co-regulate UK advertising, so creative claims need evidence before launch, not after.

Practical consent runs through a consent management platform — OneTrust, Cookiebot and Sourcepoint are common choices — and through the IAB Europe Transparency and Consent Framework, which passes the consent signal down the chain to every partner. IAB UK publishes the market guidance most UK teams work from.

Set the measurement now. Pick one primary metric and two guards, such as cost per acquisition as the primary, with viewability and frequency as guards.

Agree the budget ceiling. For example, a team paying £6,000 a month for a quarter has £18,000 to test with, excluding fees. Ask for the media fee, the platform or technology fee and any agency margin in writing before launch: they are quoted as a percentage of media spend and they change what the same £18,000 buys.

Weeks 5 to 8: run the tests

Launch with a control and two variants. A control can be your existing paid search or a broad placement set.

Build the campaign in this order: naming convention and taxonomy, audience segments, deal IDs, creative variants, then QA in the ad server before the DSP goes live. Without the taxonomy, week nine is unreadable.

Change one thing at a time. Creative, placement and bid strategy should not all move in the same week, or you learn nothing.

Use your own auction data when you set bids, and read it at the level you buy: publisher, deal ID and format. Third-party rate cards such as LinkedIn advertising costs and pricing describe a B2B auction, not the open marketplace, so they are no substitute for your DSP reporting.

Hold a weekly review on the same day. Check spend, delivery and the primary metric, then decide one action: hold, pause or scale.

Stop any line that has spent its test budget at twice the target cost per acquisition. Decide that rule in week one, when nobody is attached to the result.

Weeks 9 to 13: decide and document

Rebuild the plan around what survived. If one placement carries most of the conversions, give it more budget and a tighter frequency cap.

Write down the losers. A short note on what failed and why saves the next planner a month.

The programmatic advertising planning template gives you a structure for that write-up, including the test log and the budget lines.

Set the next quarter's objective from the evidence. If the primary metric hit target, move to scale; if it missed, narrow the audience or change the offer before adding spend.

Share a one-page summary with the client. Include spend, the primary metric, the guards and the decision.

A simple glossary

Common questions

How much should a 90-day test cost?

Enough to produce a readable result, not enough to hurt if it fails. Many English teams set a test budget between £5,000 and £20,000 for the quarter, with fees shown separately.

Can the plan cover more than one market?

It can, but each market needs its own budget and its own control. Two markets on one budget usually leaves both underfunded and neither readable.

What happens at day 90?

You make one of three calls: scale, adjust or stop. All three are valid, provided the decision follows the rules set in week one.

Do we need legal review before launch?

Get a data review if you use CRM lists or matched audiences. The ICO rights guidance is the starting point for what individuals can ask you to do with their data.