Two hundred and fifty-six weekly performance decks. The first went out on May 4 2020, the last in March 2025, and there is no gap anywhere in that run.
That file, repeated, is the reason a consulting relationship on one account lasted five years.
Most consulting engagements die quietly inside a year. Nothing dramatic happens. The work continues, the numbers are acceptable, and the client slowly stops seeing the thinking behind the invoice.
Once the thinking is invisible, only the cost is visible. Renewal conversations go bad months before anyone actually says a word about them.
I ran growth for YouMail, a consumer app for voicemail and for screening out robocalls and spam texts, as an outside consultant. The engagement had one structural weakness built into it from day one: the marketing brain lived off-site.
Nobody inside the company woke up carrying the history of what the account had tried. If I stopped writing that history down, it existed in my head and in a pile of platform exports, and nowhere else.
That is how an account's memory evaporates week to week. Six months on, somebody proposes a test that already ran and already lost, and nobody in the room can prove it.
It started as a Monday file
The first deck covered April 28 through May 4, 2020. App campaigns and Apple Search Ads were already live, creative sets were being tested against each other, and the slide worth arguing about was a landing page test.
The winning variant that spring leaned on the disinfectant angle, which was the thing everyone was thinking about in May 2020, and it beat the control on conversion rate and on cost per install, the media it takes to land the app on one phone.
Nothing about that deck felt like the start of an institution. It was a Monday file. What made it matter was the two hundred and fifty-five that came after it in the same shape.
The skeleton never changed
Five bands, same order, every week. Scoreboard, what moved and why, tests live, decisions needed, next week.
The order was the argument. The scoreboard went first because a client who has to hunt for the top-line number stops opening the file, and a client who stops opening the file is already partway out of the relationship.
The fourth band is the one client reporting almost always leaves out. Decisions needed listed the calls only the client could make: approve moving money toward iOS, kill a test that was never going to clear its bar, sign or drop a partner.
Written down, with a date attached, those items stopped the account from stalling quietly while waiting on an answer nobody had actually been asked for.
The deck was never for the client's information. It was where every decision had to defend itself every seven days.
Sameness did the rest of the work. When a given tile has sat in the same corner for two hundred weeks, anything strange in that tile is visible in about four seconds, to anyone, including people who never open an ad platform.
One rule kept the format from bloating: nothing entered the deck unless it could change a decision. Charts that existed to look thorough got cut. A deck that takes forty minutes to read gets skimmed, and a skimmed deck does the same job as no deck.
The pile turned into a database
Around year two the archive started paying off in a way I had not planned for.
Somebody would ask whether we had ever run a particular creative angle, or what happened the last time spend was pushed hard on iOS. The answer sat in a dated file with the hypothesis and the outcome already attached to it.
Five years of weekly decks is a searchable memory of every test the account ever ran. Most teams re-learn the same lesson annually.
Marketing teams lose their history constantly. People move on, tools get swapped, dashboards get rebuilt from scratch, and the reasoning behind an old call walks out with whoever made it.
A weekly artifact in a fixed shape is the cheapest defense against that I know of. It also ended arguments, by making the record checkable while the argument was still happening.
What compounding looked like in the numbers
FY2024 is the year the archive can speak for cleanly. The account deployed about $550K in paid media at a blended cost per install just over $2, about $5 per registration, and about $120 per paying subscription.
Those figures are not spectacular in isolation. They are what an account looks like after four years of small corrections that nobody lost the thread on.
The record also carries the costs that went the wrong way, which is the harder half of its value. In 2022 the account spent roughly $510K and paid about $56 for each new subscriber. By FY2024, on comparable spend, that cost had reached about $120.
In a weekly series you watch that drift arrive. In an annual review you find out about it after the money is spent. Watching a cost double over two years is uncomfortable; discovering it at the end of year two is worse.
Across the relationship, paid acquisition climbed from nothing to roughly 15% of the new business coming in. Every step of that climb came out of a call that had been argued through in one of these files first.
Rhythm is the actual product
The hard decks were the quiet weeks. Nothing moved, no test concluded, and the honest content was the position we held, the reason we held it, and what lands next.
Those weeks are the real test of the format, because there is no result to hide behind and the only thing on offer is your reasoning.
A consulting engagement survives on rhythm. The week you skip the report is the week the client starts wondering what they pay for.
Five years of Mondays is about 250 chances to look like the person holding the account together. The deck is how you spend those chances deliberately instead of hoping the results speak for themselves.
Run inside a team, this discipline becomes the weekly growth review, a live meeting where colleagues decide together. The deck is the client-facing counterpart, the version that keeps an outside operator's reasoning legible to people who were never in the room.
Most of what these files reported on has been written up on its own: the acquisition and conversion rebuild, the search program built on fifty dollars a day, and the answer when Apple shipped call screening into iOS. Each of those was a slide before it was a project.
Build the habit on your own account
You do not need five years to get the benefit. You need a skeleton you refuse to redesign, and a day of the week you refuse to move.
- Fix the five bands and refuse to redesign them. Comparability is worth more than a fresh layout.
- Put the scoreboard on the first screen, with the prior week and prior year beside it, so nobody has to hunt.
- Name a cause next to every delta, and write "cause unknown" when that is the truth. Unknowns are what next week's tests are for.
- Record each live test with the hypothesis it was launched on, so a result cannot be reinterpreted after the fact.
- End with the decisions you need and the dates you need them by. That band is what makes the file an operating tool.
- Send it on the quiet weeks too. The unbroken run is the part the client actually notices.
Do that for a year and you have something most accounts never build: a record of your own reasoning that anyone can search. Do it for five and the record starts making decisions cheaper, because the expensive question has usually been answered already.
If your reporting reads like an invoice with charts attached, the fix is usually the skeleton rather than the effort. This is the kind of operating rhythm I install. Let's talk.