Marketing Reporting in a Multi-Source World
Paid platforms report one number. GA4 reports another. Your CRM reports a third. Here is how to build a marketing report that reconciles them without pretending the reconciliation is clean.
The reconciliation problem
Any marketing report that claims to show "one number" across paid, organic, email, and direct is either lying or ignorant. Platforms disagree. GA4 disagrees with your CRM. Your CRM disagrees with Stripe. This is not a failure of tooling. It is a property of the world.
A good marketing report does not hide the disagreement. It shows the numbers side by side, labels them with the attribution model that produced them, and picks one as the report-of-record for that period.
The canonical monthly marketing report
1. Executive summary
Three sentences. One win, one miss, one thing you are testing next.
2. North star + inputs
Pipeline, qualified leads, MRR attributed to marketing, whatever your company uses. Underneath, the five-to-seven input metrics that drive it. MoM and YoY deltas.
3. Channel breakdown
Per-channel spend, CAC, CPC, CTR, conversion rate, and ROAS. One row per channel. A good report ranks channels by net new customers, not by spend.
4. Campaign highlights
Three campaigns: best performer, worst performer, most surprising. Name them. Show the creative where possible.
5. Attribution appendix
Last-click view, data-driven view, and an honest note on the delta. This is where good marketers separate from mediocre ones.
6. Experiments and learnings
What you tested this month. What you learned. What you are testing next.
7. Asks
Budget asks, hiring asks, tooling asks. Made in the open.
Attribution: pick a stance and hold it
Attribution is the most over-debated topic in marketing. The practical answer: pick one model as your report-of-record, publish it consistently, and annotate when you see material divergence from platform-reported numbers. The model matters less than the consistency.
- Last non-direct click - stable, easy to explain, biased toward bottom of funnel.
- Data-driven (GA4) - better at crediting upper-funnel channels, requires volume.
- First-touch - good for brand-investment arguments, unfair to direct response.
- Position-based (40/20/40) - opinionated, reasonable, rarely worth the complexity.
The cohort layer
CAC averaged across a month is noise. CAC by acquisition month, tracked forward over payback period, is signal. A good report shows spend payback cohorts for at least the trailing six acquisition months.
Common failure modes
- Platform-inflated numbers. Reporting Meta-attributed conversions as "conversions from Meta" without context. Executives who realise the trick lose trust in everything else.
- CAC without time horizon. Monthly CAC is a rolling average. Payback period is the real answer.
- Vanity over actionability. Impressions and reach belong in the appendix, not the headline.
- Channel silos. Reporting each channel as if it exists alone, ignoring the interaction effects.
Automation and the marketing report
A monthly marketing report typically pulls from 8-15 sources: Google Ads, Meta Ads, LinkedIn Ads, GA4, Search Console, HubSpot or Salesforce, Stripe, Mailchimp, Segment, and whatever else the team runs. Donum connects to all of them, pulls the relevant slices, writes the narrative sections, and delivers the report on the first of every month. The marketing leader's job becomes editing, not assembling.
Frequently asked questions
Why do Meta Ads and GA4 show different conversion numbers?
They measure different things. Meta counts conversions attributed inside Meta (usually 7-day-click, 1-day-view by default). GA4 counts conversions in the GA4 data model using its own attribution. Neither is wrong; they are answering different questions. A good report shows both with clear labels.
Last-click or data-driven attribution?
For day-to-day channel reporting, data-driven is better when your volume supports it (roughly 3,000+ conversions per month per property). For channel health checks at smaller scale, last-click is more stable, even though it is less accurate. Do not mix models within the same report.
How often should a marketing team report?
Weekly snapshot for the team (lightweight, focused on anomalies). Monthly report for leadership (full structure, narrative, recommendations). Quarterly business review for executives (trends, strategy, resource allocation).
What is the right number of KPIs on a marketing report?
Seven, give or take two. More than ten and stakeholders stop reading. Fewer than five and you are probably hiding something. Pick the north star plus five to seven input metrics that drive it.