The principle

The primary metric should sit close to the business.

Impressions, clicks and CTR help explain campaign behavior. They should not be the final point of reporting when there is a measure closer to revenue, margin, qualified opportunity or retention.

The farther the metric is from revenue, the greater the risk of optimizing an indicator while making the system worse.

A simple structure

Four layers organize the analysis.

01

Delivery

Reach, frequency, impressions and cost show whether the campaign was able to compete and distribute.

02

Response

Clicks, views, installs and forms explain the response to the message and experience.

03

Quality

Activation, opportunity, proposal, trial and purchase reveal whether the acquired profile makes sense.

04

Economics

Revenue, margin, CAC, payback, LTV and retention show whether the operation is sustainable.

One layer does not replace another. The problem begins when the operation tries to answer an economic question using only a delivery metric.

Less false certainty

Attribution does not need to be perfect to be useful.

Platforms, analytics and CRM record different parts of the journey. Instead of declaring one source of truth, the analysis can triangulate events, UTMs, offline conversions and aggregate behavior to support a decision.

Campaign signalPlatform and UTMs
+
Business signalCRM, sale and retention

What increases confidence

  • A UTM standard applied without meaningful exceptions.
  • Events with clear names, definitions and owners.
  • Importing opportunity or sale events when the cycle happens outside the website.
  • Trend and cohort analysis, not only point-in-time conversion.
  • Periodic reconciliation across platform, analytics, CRM and finance.

Reporting that produces action

Each window allows for a different kind of decision.

Seven days help identify anomalies and make small adjustments. A month reveals enough trend for larger decisions. Cohorts and LTV require longer windows. The cadence needs to respect the maturity time of the business.

The report needs to answer

  • What changed and at which stage of the operation.
  • Which hypothesis explains the change with the available evidence.
  • What is under the team’s control and what depends on another area.
  • Which action is safe now and what still deserves observation.

Operational evidence

Changing what is measured changes what is decided.

E-commerceAudrey

ROAS and CAC analyzed by channel and period to support more than R$20 million in revenue across the partnership.

B2B high ticketPonce

The operation moved beyond isolated volume and followed the journey through opportunity and sale.

Lead generationBiovis

More than 22,000 leads with qualification sustained above 60% and tracking of the real outcome.

Common questions

Good measurement reduces doubt. It does not promise absolute certainty.

Are ROAS and CPL bad metrics?

No. They are incomplete when shown without segmentation and without a connection to margin, quality or revenue.

Which dashboard should we use?

The tool comes after the definition: question, event, source, calculation and cadence. A beautiful dashboard does not fix poorly defined data.

How do we measure offline sales?

With UTMs, lead identifiers and CRM feedback. The level of automation depends on the structure, but the cycle can start simply.

Which metrics matter first?

CAC, value generated, payback time and one measure of quality or retention. Everything else should help explain those four.