Flagship programme
Lift Studio: Designing Incrementality Tests
An eight-week studio for people who already own a media or product number and need conversion lift measurement that survives challenge. Taught live from Worminster, with recorded catch-up for GB time zones.
What you will be able to do
- State a lift claim in a sentence a finance partner can reject or accept.
- Choose a unit of randomisation you can actually implement in your stack.
- Size a test against a media calendar rather than an infinite sample.
- Spot contamination from frequency caps, CRM, and retailer networks.
- Write a results memo that includes the residual uncertainty on purpose.
Who it is for
Analysts, media leads, and product managers who already ship weekly reporting. We do not teach spreadsheets from scratch. If you have never seen a confidence interval, start with the Holdout Architecture short course instead.
Modules
1. The claim
Turning “the campaign worked” into a falsifiable incrementality statement, including the population you are willing to leave untreated.
2. Units and interference
Users, households, cookies, stores, and postcode sectors — and why the wrong unit quietly inflates lift.
3. Holdouts that still exist in week six
Leakage maps for paid social, search, and CRM. This is the week most briefs get rewritten.
4. Power without theatre
Minimum detectable lift against spend, seasonality, and the political cost of a long test.
5. Geo designs
Matched markets for GB, synthetic controls as a check, and weather as a covariate rather than an excuse.
6. Product flags
Exposure logging, network effects, and stopping rules when a holdout starts to look unethical.
7. When MMM argues back
A protocol for the week marketing mix and lift refuse to agree, without declaring a winner by seniority.
8. The finance memo
A marked assignment: two pages, one appendix, and a list of things you still cannot know.
Tutor
Helen Cartwright
Helen ran measurement for a UK grocer’s media function for six years, then a travel subscription product. She still marks every Lift Studio memo herself. She will tell you, early, if your proposed holdout is theatre.
Fee
The published fee is £1,840 for one seat in the live cohort, including critique slots and recordings for eight weeks. Employers may pay by invoice; we do not take card payments on this site. See fees for other formats and refunds for eligibility.
A limitation you should know
Lift Studio will not give you a vendor-certified badge, and it will not make a two-week burst campaign statistically decisive. If your organisation refuses any untreated group, the methods here stay theoretical. We say that in week one so you can leave without wasting a quarter.
Questions we hear before enrolment
Do I need Python?
No. Examples use spreadsheets and a little SQL. People who prefer notebooks are welcome to use them; we mark the design, not the stack.
Can a whole team join?
Yes. The In-house Measurement Desk on the fees page is the format for that. Lift Studio seats are individual so critiques stay specific.
What if our data is a mess?
Then you are in the typical case. Module three spends more time on what you cannot log than on perfect event schemas.
Is there a guaranteed lift number at the end?
No, and treating the course as a way to manufacture a positive result is a reason we would refuse a seat. Conversion lift measurement includes the possibility of a null.
Notes from people who sat the studio
The finance memo assignment was marked harder than our internal comms team ever marks us. I reused the structure the following Monday. I would have liked a second marked draft; one pass still left some of my hedging clumsy.
Anonymous client in subscription media