Measurement
How to brief a privacy first measurement pilot without a cookie banner
Scope a six to eight week privacy first measurement pilot with named owners, an evidence pack, stop conditions and no change to your cookie banner.
What to take away
- One decision, one owner, one deadlinewrite these three lines before you touch any tag.
- Use first party events already collected under your existing notice, so no new cookie banner is required.
- Fix the evidence pack and the stop conditions in the brief, not after the results arrive.
- Cap the pilot at six to eight weeks, one channel and one funnel stage.
- Book the readout with the budget holder on the day the brief is signed.
Write the decision before the brief
A privacy first measurement pilot brief is one page. It starts with a decision the business must make and cannot answer today. Good examples include whether paid social earns its budget, or whether email drives repeat purchase.
Bad examples are open questions such as understanding customers better. Those cannot be tested and cannot be stopped. Write the decision as a sentence with a deadline attached, then add the name of the person who acts on the answer.
What the one page brief contains
- Decision and deadline. State the choice, the date it must be made and the budget it controls.
- Owner and deputy. Name one person who signs off findings.
- Data sources. List the first party events, CRM fields and platform exports in scope.
- Method. Choose an incrementality test, marketing mix modelling or a consented event study.
- Evidence pack. Define the charts, tables and ranges the readout must include.
- Stop conditions. Write the results that end the pilot early.
- Budget and time. Set the maximum spend, the end date and the review date.
A brief template in table form
Use this table as the spine of the document. Every cell needs a name or a number, not a description.
| Brief section | What to write | Sign off |
|---|---|---|
| Decision | One choice with a deadline | Budget holder |
| Owner | One named person plus a deputy | Head of marketing |
| Data | First party events, CRM, platform exports | Data protection lead |
| Method | Incrementality test, MMM or event study | Analyst |
| Evidence | Charts, ranges and sample sizes | Owner |
| Stop rule | Thresholds that end the pilot | Owner and finance |
Example: a six week test brief
Suppose a homeware brand wants to know whether paid social drove first purchases. The brief names the ecommerce director as owner and sets a six week window.
The team uses consented first party purchase events and a holdout group. No new banner is needed because nothing changes in what they collect or why.
The readout compares incremental orders per thousand pounds against a two week baseline. The pilot stops early if the holdout gap falls below the agreed threshold.
Evidence to collect and stop conditions
Collect the smallest set of numbers that can change the decision. A checklist keeps the analyst honest.
- Incremental conversions and revenue per channel, with ranges.
- Sample size and the number of days observed.
- Cost per incremental order against the agreed ceiling.
- A written note on what the result cannot prove.
Stop the pilot if the holdout group cannot be kept clean, if event volumes fall below the agreed floor, or if legal review flags the data flow. Agree all three thresholds in writing before week one.
Choosing a method that fits the decision
An incrementality test suits one channel with enough volume to split. Marketing mix modelling suits total budget questions across channels over longer periods, and the Wikipedia article on marketing mix modeling covers its inputs and history.
Owners, exports and the readout
Name the decision owner, the analyst and the data protection lead in the brief. A missing name is the most common reason a pilot drifts. Platform exports describe delivery, not incremental effect, so treat them as supporting evidence.
UK cookie consent and electronic marketing rules sit in the Privacy and Electronic Communications Regulations 2003, with further detail in the ICO guidance on direct marketing and privacy. The Data Protection Act 2018 supplements the UK GDPR and covers the personal data a pilot touches.
Social teams that need a working method before the pilot starts can read how to measure social media ROI without third-party cookies.
Once a pilot proves one decision, the next question is where similar briefs come from. A short read on five buyer decisions that create privacy first measurement work in England shows the buying triggers to watch.
Common questions
How long should a pilot run?
Six to eight weeks is enough to see a holdout gap in most mid-sized UK accounts. Anything shorter risks noise, and anything longer delays the budget decision.
Does the pilot need a new cookie banner?
No, if you only analyse first party events already collected under your current notice. A new banner is required when you change what you collect, why you collect it or who receives it.
What should stop a pilot early?
Three triggers end it cleanly: a contaminated holdout group, event volumes below the agreed floor, or a legal review that flags the data flow. Write all three thresholds into the brief before week one.
