How it works

Six stages. Each one ends in something you can hand over.

Diagnose, Decide, Build, Launch, Improve, Reconcile. Run them in order for a new plan, or jump straight to the one matching what is on your desk today.

The last stage is the one nobody else sells. Every media product stops at reporting. The work that actually consumes a planner's month is what comes after it.

Diagnose

A brief is rarely wrong. It is usually incomplete in ways that stay invisible until you are three days into building and discover the conversion event does not exist. Thirteen flags, each with the question to ask, the decision that depends on it, the assumption you can hold temporarily, and the risk to record if you proceed without it.

InputThe brief, as received

ToolBrief Intake tab, 22 fields

OutputA short list of questions, blocking ones marked

Common mistakeAnswering the brief instead of diagnosing it

13 flagsGap counterFeeds QA Notes

Decide

Markets, platforms, funnel and objective, in that order. A market gets budget when it can convert, not when it looks large. And the most useful output is the one everybody skips: the written reason for every platform you left off, because defending an exclusion is harder than defending an inclusion and you will be asked.

InputA diagnosed brief

ToolMarket and Platform Scoring

OutputA weighted split with a reason per line

Common mistakeAn equal split, which is what you produce when nobody would defend a decision

4 market playbooks8 platform cardsExclusion reasonsFunnel decision tree

Build

Budget allocation with a fragmentation flag per line, a forecast that runs spend through to contribution margin, a scenario planner, and the affordable cost per lead derived from the client's own economics. Four lines of arithmetic, no benchmark of any kind, which is exactly what makes the answer impossible to argue with.

InputMarket and platform decisions

ToolMaster planner, Commercial calculators

OutputA plan, a forecast, a conservative case, a client export

Common mistakeAccepting a KPI instead of deriving one

Fragmentation checkAffordable CPLBreakeven ROASFunnel sensitivityScenarios

Launch

Measurement first, because everything decided here is expensive to change once money is moving. Then forty-seven checks across account, campaign, creative, measurement and governance. Then a real test conversion, by a human, today, verified in all four systems.

InputAn approved plan

ToolActivation QA, Launch Tracker

OutputA signed-off tracker with evidence

Common mistakeA check marked pass with no evidence recorded

47 checksGo / no-go verdict19-item measurement checklistWhatsApp handoff

Improve

Classify before you change anything. Delivery, attention, intent, conversion or quality. Almost all wasted optimisation effort is a competent fix aimed at the wrong layer, because bid strategy is the lever nearest to hand when the real problem is a page that takes nine seconds on a phone.

InputA campaign that is not working

ToolOptimisation diagnosis, Reporting system

OutputOne change, a read date, client-safe wording

Common mistakeFour changes in one week, then reading the result

5 problem classesWhat not to change yetEvidence required12 commentary situations

Reconcile

Approved, PO, actual, accrued, invoiced, paid. Six numbers that should agree and will not. A documented source-of-truth hierarchy settles it, and one rule resolves most of what this stage involves: an invoice never overrides an actual, because the platform is the authority on its own spend.

InputA month that happened

ToolPacing and reconciliation

OutputA variance report and a list of exposures

Common mistakeDoing it at month end, from memory

Source-of-truth hierarchy7 automatic flagsOutstanding exposurePO expiry tracking

The loop

Reconcile feeds back into Diagnose.

What the invoice actually cost, what the CRM actually recorded, and what the contactable rate actually was become the assumptions on the next plan, with a source and a date attached.

That is the compounding part. The first plan runs on the best rates you can find. The fourth runs on your own, and by then nothing in this system is telling you anything you did not measure yourself.

Planning Louli reality check

Take the monthly budget, divide by markets, then platforms, then funnel stages. If the cell that comes out cannot clear minimum viable daily spend, you do not have too little budget. You have too many lines. That check takes ninety seconds and it is almost never done before the plan goes out.

See all six stages on one campaign What's inside