Reports exist to change decisions
The test of any report is what it caused. If the weekly report was read, discussed and filed, and the budget, bids, creative and targets were the same on Monday as they were on Friday, the report was decoration. That is not always wrong, because sometimes the right decision is to change nothing, but it should be a decision rather than a default.
Reports drift into decoration for predictable reasons. They are built from what the platform exports, so they lead with impressions and click-through rates. They report each channel on its own AttributionAssigning conversion credit to touchpoints., so the totals add up to more than the business received. They show what happened without saying what was planned, so nobody can tell whether the result is good. And they carry no record of the previous decision, so each review starts from nothing. Commentary written as what happened, so what and now what makes every section state the result, its meaning and the change it leads to.
The measurement plan framework is the corrective. It defines, before any report is built, what is measured, by which rule and source, reviewed by whom and when, and which decision each measure informs. A report designed from the measurement plan carries only what someone will act on. The optimisation hub places reporting at the end of the test-and-learn loop, where it closes each turn.
Cadence: daily checks, weekly optimisation, monthly review
Different decisions need different rhythms. Mixing them, by reviewing strategy daily or catching broken tracking monthly, is the main source of wasted reporting effort.
| Cadence | Purpose | What is looked at | Decisions it can make | Who |
|---|---|---|---|---|
| Daily check | Catch breakages | Spend against pacing, conversions recorded, disapprovals, tracking alerts | Fix, pause, escalate | Channel owner |
| Weekly optimisation | Manage within channels | Cost per result against target by campaign, creative performance, test status | Shift budget within a channel, refresh creative, start or stop tests | Channel owners and lead |
| Monthly review | Manage across channels | Plan against actual in the business's data, blended cost, reconciliation, marginal view, decision log | Move budget between channels, change targets, commission tests | Marketing lead with finance |
| Quarterly review | Re-plan | Incrementality results, modelling, economics inputs | Re-set roles, budgets and targets | Leadership |
The daily check should take minutes and exists to catch problems such as a conversion tag that stopped firing or a campaign that overspent. It is not an optimisation session; acting on one day's cost per acquisition (CPACost of each conversion.) mostly reacts to noise, and the guide to reading small numbers in marketing reports shows how to tell a campaign's weak week from ordinary variation when its counts are low. The weekly session is where most in-channel work happens. The monthly review is where money moves between channels, and it needs data that has settled through attribution windows. The monthly paid media review checklist is that meeting's agenda in tickable form, from the tracking check to the decision log.
The widget below lays out a cadence for your team so that each meeting has a clear owner and decision scope.
Reporting cadence planner
The report structure: plan, efficiency, volume, marginal
A monthly performance report that drives decisions usually has four sections, in this order.
Plan against actual. The outcome the business cares about, measured in its own systems, against what the plan said: new customers, revenue, qualified leads, and the blended cost of each. This is the headline and it should fit in a few lines. If the headline is on plan, the rest of the report is context; if it is off plan, the rest is diagnosis.
Efficiency. Cost per acquisition and return on ad spend (ROASRevenue generated per unit of ad spend.) by channel and major campaign, against each one's target, with the attribution source labelled. The ROAS calculator is useful for checking any figure that looks implausible.
Volume. Results and spend by channel. Efficiency without volume is misleading: a channel can hit its target by shrinking. The plan said how much each channel should deliver, and volume shows whether it did.
Marginal view. What the last increment of spend bought in each channel, from budget steps, scaling tests or modelling. This is the section that justifies moving money, because it shows where the next pound would work hardest. The testing and scaling guide explains how marginal cost is estimated.
The monthly marketing report template provides a layout that follows this order. At the end of a quarter, the quarterly marketing review template reads three months of results through their drivers and puts the decisions requested at the top.
Where paid social is a large share of spend, the creative section of the pack can follow the tutorial on reporting on Meta creative, which groups every ad by concept, hook and format and ends in the next round of briefs.
For TikTok, where paid and organic results overlap, the tutorial on building a monthly TikTok report joins Ads Manager, the account's analytics and the business's own records so that Spark Ad engagement and view-through conversions are counted once.
Reconciling platform and analytics numbers
Platform figures, analytics figures and the business's own records will never agree, and they are not meant to. Each platform counts conversions it touched under its own attribution window and model; analytics applies one model across all channels, using tagged sessions; the business records orders and leads without knowing where they came from. The Google Analytics help centre documents how its attribution settings and reporting identity work, and each ad platform documents its own. The Media Rating Council sets standards for how impressions and other measures are counted, which explains part of why figures differ even before conversion is considered.
Reconciliation is not making the numbers match. It is knowing, each month, the size of the gap between each source and the business's records, and noticing when the gap moves. A steady gap is a property of the measurement; a sudden change is a fault, such as broken tagging, missing UTMURL tags identifying traffic source. parameters, a changed attribution setting or a duplicated event. The platform vs analytics discrepancies guide covers the diagnosis, and the attribution approach guide covers which model the report should use for cross-channel comparison. The attribution models framework sets out the options.
A simple reconciliation table belongs in every monthly report: business-recorded results, analytics-attributed results, and each platform's reported results, with the ratio of each to the business figure and the same ratio last month.
When the platform columns come from a warehouse rather than the interfaces, the guide to loading ad platform data into a warehouse explains why recent days have to be reloaded as late conversions arrive, or the latest month will always look worse than it was.
Incrementality and modelling for the bigger questions
Dashboards answer what happened under a set of attribution rules. They cannot answer whether a channel caused the results it was credited with, or what would happen if its budget doubled or halved. Those are the questions that decide budgets, and they need different tools.
IncrementalityConversions caused by marketing. tests answer the causal question for one channel at a time by withholding ads from a random group or region and measuring the difference. They are precise for the channel tested and the period they ran, and they need planning, budget and patience.
Marketing mix modelling (MMMStatistical model of channel impact.) answers the allocation question across all channels at once, estimating each one's contribution and diminishing returns from historical spend and outcomes. It needs a long, varied history and skill to build, and its estimates are only as good as that history. The marketing mix modelling basics guide covers when it is worth building.
In a mature programme the three work together: dashboards for weekly management, tests to calibrate the dashboards' claims for each channel, and modelling to set the budget split. The monthly report should state when each channel was last tested and what the result was. When a campaign ends, the campaign wrap-up checklist makes sure that question is asked of it before the results are presented.
The decision log
The decision log is the cheapest and most valuable part of performance reporting. It is a simple running record: the date, the decision, the evidence behind it, the expected effect, and, at a later review, the actual effect. It turns a series of meetings into an institutional memory.
Without it, teams repeat tests they ran a year ago, reverse decisions without knowing why they were made, and lose everything when someone leaves. With it, a new team member can read the account's history in an hour, and every review starts by checking whether last month's decisions did what they were meant to.
Keep entries short and specific. A useful entry says what changed, why, what was expected and when it will be checked. The log sits at the end of the monthly report, and its last few entries are the first thing the next review reads. The marketing decision log template lays the log out as columns, adding the reversibility, guardrail and outside events of each decision, with a filled example.
A B2B software company rebuilds its monthly review around decisions
Ledgerline, an accounting software company, produced a monthly report of 34 slides, mostly platform screenshots, and its reviews rarely changed the £180,000 monthly budget. The rebuilt report opened with plan against actual from the CRM: 412 qualified trials against a plan of 450, at a blended cost of £437 against a target of £400. Efficiency and volume followed by channel, then a reconciliation table showing Google claiming 1.1 times and LinkedIn 1.6 times the trials recorded in the CRM. The marginal view showed the last £20,000 on LinkedIn buying trials at £780. The review moved £15,000 to search and commissioned a geographic holdout on LinkedIn, and both decisions went into a new log with a check date two months later.
Common mistakes
- Leading with platform metrics. Impressions and click-through rates invite a discussion of activity; lead with plan against actual in the business's own data.
- Summing platform-reported conversions. Each platform claims overlapping results; the sum overstates what the business received.
- Optimising on daily figures. Daily checks catch breakages; acting on one day's cost per result reacts to noise.
- Efficiency without volume. A channel can meet its target by shrinking; report results and spend alongside cost.
- No reconciliation. Without tracking the gap between sources, a broken tag can run for weeks before anyone notices.
- Treating dashboards as causal. Attribution says who was credited, not who caused the result; budgets need tests and modelling.
- No decision log. Reviews without a record repeat old tests and reverse good decisions for forgotten reasons.
