If you ask your agency for a monthly report and the first page leads with impressions, reach, CPM, and a frequency chart, you are looking at a 2008 marketing report. The numbers are real. They are also nearly useless for any decision a CMO actually has to make.

The four metrics below replace most of what your current agency dashboard shows. Together, they answer the question that matters: is marketing producing more revenue per dollar than it would have if you'd done nothing, and is that ratio trending up?

1. Blended CAC

The cost to acquire a customer, calculated across all marketing spend — not the channel-specific CACs that platforms report. Sum every dollar spent on paid media, content, agency fees, tech, and influencers in a window. Divide by net new customers acquired in the same window. That number is your true cost of customer acquisition.

Why blended? Because channel-specific CACs are over-credited by every platform that reports them. Sum-of-platforms always exceeds true blended CAC by 15–60% depending on your channel mix. The blended number is the one that ties to your cash flow.

The ask: monthly blended CAC, plotted against your contribution margin per customer (see metric 3). The ratio of these two — and its trajectory — is the single most important business signal in marketing.

2. Contribution margin on marketing-attributed revenue

Revenue alone is misleading. Spending more to drive low-margin revenue is a way to grow yourself into a worse business. The metric you actually want is contribution margin — revenue minus variable costs (COGS, payment processing, shipping, refunds, and the marketing cost itself).

Most agencies cannot produce this number because they don't have the cost data. Demand it anyway. The work to produce it forces a conversation about which segments and channels are actually profitable rather than just driving topline.

The ask: monthly contribution margin, segmented by acquisition channel and customer cohort. If your highest-revenue channel produces below-average contribution margin, you may be funding a channel that's making your unit economics worse.

3. ROAS by audience cohort

Aggregate ROAS hides the answer. A 3.5× blended ROAS may mean every audience returns 3.5×, or it may mean two audiences return 6× and three return 0.8× and the average looks fine. Those two situations require completely different actions.

The ask: ROAS broken out by the five to ten audience cohorts that matter for your category. For hospitality this is usually combinations of geo, booking-window, party-type, and price-tier. For retail it's cohort + product-category + repeat-vs-new. For B2B it's industry + company-size + product-line.

If your agency cannot produce cohorted ROAS, they cannot tell you which audiences are subsidising the others. That blindness is where most budget waste lives.

4. Creative win-rate

Of every creative variant produced and tested in a window, what percentage met the "kept-in-rotation" threshold (typically 1.2× the campaign-average ROAS or higher)?

This number tells you whether your creative engine is working. A healthy AI-native creative operation should be producing variants at high volume with a 20–35% win-rate. A traditional agency producing 6 variants per campaign and rotating them all regardless of performance has an undefined win-rate, which is often worse than a low one — it means there's no measurement happening.

The ask: number of creative variants tested per channel per month, win-rate against the kept-in-rotation threshold, and the win-rate trajectory over the last six months. Trajectory matters more than absolute number; a flat win-rate over six months means the creative engine has plateaued and needs intervention.

What to stop asking for

Reciprocally, you can stop asking for and stop reading:

  • Total impressions and reach (vanity, not directional)
  • Click-through rate without conversion context (optimised on the wrong axis)
  • Engagement rates on social posts (correlation with revenue is approximately zero)
  • Cost-per-mille (CPM) as a primary metric (a cost ratio, not an outcome)
  • Sentiment dashboards (qualitative, often unactionable, prone to being gamed)
  • Share-of-voice charts (a competitive metric, not a performance metric)

None of these are inherently bad. Used as inputs to the four metrics above, some are useful. Reported as headline numbers, they create the illusion of insight without the substance.

The follow-up question

If your agency can produce the four metrics above, the follow-up question is this: what would you do differently if blended CAC went up by 30% next month, and what would you do differently if it went down by 30%?

If the answers are precise — specific channels to throttle or expand, specific audiences to deprioritise or scale, specific creative archetypes to retire or amplify — you have an agency operating on the right metrics. If the answers are vague, you have an agency that has the dashboards but isn't using them as decision tools.

That distinction is the difference between marketing reporting and marketing strategy.