Flagship course
The Install Graph Method
A six-module lab for people who already ship paid UA and still cannot say, without flinching, which campaign produced Tuesday’s UK iOS installs.
What you leave with
Not a certified MMP operator badge. You leave with an install graph for one live product, a written match policy (lookback, view-through rules, SKAN fallback), and a one-page board note that names the uncertainty instead of burying it in a footnote.
Homework uses your redacted exports. If your legal team will not release even hashed campaign IDs, we issue a GB fintech sample set — clearly labelled as a stand-in, never as your truth.
- Draw campaign → click/view → store session → install as a directed graph.
- Reconcile ads manager counts with store and MMP counts without “plugging the gap”.
- Document when last-click is allowed as a residual, not as a religion.
- Present a limitation: SKAN sparsity in the first 48 hours after a creative swap.
Modules
Objects on the graph
Campaigns, ad sets, creatives, store listing experiments, and organic search as first-class nodes. We refuse to treat “network” as a single blob.
Clocks and lookbacks
Click-through versus view-through, GB evening traffic, and why a 24-hour window lies differently on payday Fridays.
Privacy windows
ATT prompts, SKAN conversion values, and what a GB publisher can still join without a device graph fantasy.
Creative versus listing
Diagnostics when the asset wins the tap and the product page loses the install. Includes the memo template from the studio day.
Reconciliation theatre
How finance, the agency, and the MMP can all be “right” on different grains — and how to write a single table they will all initial.
The match policy
You ship a two-page policy and a live board. Tutors mark the policy harder than the dashboard polish.
Tutor
Helen Okoye
Former in-house UA lead for a London marketplace, now teaching the Install Graph Method. Helen still reviews every match policy in the Lab Seat. She will send yours back if the SKAN section is a screenshot with no prose.
Questions we actually get
Do you teach a specific MMP?
No. We use whichever export you bring. The method is the graph and the policy, not a vendor certification.
What if our iOS volume is too thin for SKAN?
That is a real limitation, not a homework fail. Module three requires you to state the minimum events before a conversion value is trustworthy. If you are below it, the honest output is “do not steer creative on SKAN this week.”
Is the fee refundable?
See refunds. Desk materials already issued are treated differently from a seat cancelled before the first live session.
From people who sat the lab
“Module five stopped our weekly ‘whose number is it’ argument. The graph still looks ugly when a brand campaign dumps view-throughs on a Sunday.”
Rated 4.2 by the last Lab Seat on a five-point tutor survey — we publish the score because it is not a round 5.0.