Diagnosis and plan
- Measurement diagnosis of what exists today
- Definition of the indicators that matter for the company’s objectives
- Measurement plan, with a written definition for each indicator
- Organisation of data sources
IntelligenceMarketing intelligence
Data does not make decisions. Well-informed people do.
We organise indicators, interpret results and turn information into priorities, corrections and opportunities.
The diagnosis starts from the decisions that need making, not from the tools in use.
The problem
Information is not what is missing. Conclusions are.
Most companies that say they have no data have too much of it — scattered across platforms that contradict each other and that nobody reconciles.
What is usually going on
- Plenty of reports circulating and few decisions coming out of them.
- Each platform reporting a different number for the same conversion.
- No consolidated view across channels, site and sales.
- Campaigns judged by clicks because that is what is available.
- No reading at all on the quality of incoming leads.
- Marketing assessed with no connection to revenue.
- Data presented without the context that makes it interpretable.
- Carefully built dashboards nobody opens after the first month.
- No stated success criteria — every meeting redefines what "good" means.
- Decisions taken on the opinion of whoever speaks loudest.
- No way to locate waste, even while suspecting it exists.
What it costs
- Budget kept in a poor channel because there was never a fair comparison.
- Good initiatives shut down too early, on a mistaken reading.
- Recurring arguments about numbers instead of discussions about decisions.
- Marketing treated as an expense, because it cannot demonstrate contribution.
A report is not a result. A decision is.
How we think
The dashboard is not the product
A dashboard is not the final product. It is an interface for understanding what needs to be done. A beautiful dashboard nobody consults cost the same as a useful one and produced no decisions.
That is why the work starts at the end: which decisions does this company need to take, how often, and who takes them. From there it becomes possible to say which indicators matter — and, more importantly, which can be ignored without loss. Measuring everything is the most elegant way of looking at nothing.
The third point is precision. No marketing measurement is exact: blockers, consent, multiple devices and attribution windows guarantee that platforms will always disagree. The objective is not the perfect number, it is a reading consistent enough to compare periods and decide.
- Start from the decision
- An indicator that changes no decision can leave the report without anyone missing it.
- Consistency before exactness
- Better an imperfect number measured the same way every time than an exact one whose definition changes each quarter.
- The reader has to understand it alone
- A report that only makes sense with someone explaining alongside is not a report: it is a presentation.
What it is
The layer that turns measurement into priority
Marketing intelligence is the work of defining what to measure, organising the sources, interpreting what the numbers show, and converting that into a recommended action — a correction, a reallocation or a bet.
The technical part — events, conversions, dashboards, reconciled sources — is a condition, not the deliverable. It exists so that the reading is possible. What the company is buying is the reading.
That includes saying what cannot be claimed. A good analysis separates what the data supports, what it merely suggests and what remains a hypothesis — instead of giving everything the same degree of certainty in order to sound conclusive.
Scope
What the area covers
The composition depends on what is already measured, the tools in use and the decisions at stake.
Implementation
- Events
- Conversions
- Dashboards inside the tools already in use
- Reconciliation between platforms where numbers diverge
Analysis
- Funnel analysis
- Lead quality analysis
- Channel analysis
- Campaign analysis
- CAC, LTV, ROAS, revenue and retention, according to available data
Decision
- Hypothesis formulation
- Experiment design
- Interpreted executive reports
- Recommendations with stated priority
- A follow-up routine with the people who decide
What is not included
- Development of custom connectors
- Bespoke data engineering
- Building a proprietary BI platform
- Complex technology integration as the core service
Where the need involves software, custom connectors or engineering, it is assessed together with Centriu. We do not promise that within the Arte com Pimenta offering without that validation.
Not every project uses every indicator. Demanding CAC, LTV and margin from an operation that does not yet record sale origin would create work before creating usefulness.
Vocabulary
What each indicator answers
Before comparing numbers, it is necessary to agree on what each one means. This is the definition we use.
- CPL
- What each generated lead costs, by channel and by campaign.
- CPA
- What each action defined as a relevant conversion costs.
- CAC
- What it costs to win an actual customer, counting the investment involved.
- ROAS
- Return on media investment, within the scope that was measured.
- LTV
- What a customer represents across the relationship, not only at first purchase.
- Conversion
- Where the journey advances and where it stalls, stage by stage.
- Revenue
- How much turnover traces back to marketing.
- Margin
- When the company shares the figure, it shows whether the sale is profitable.
- Retention
- Whether the base stays — which changes the acquisition maths entirely.
- Lead quality
- Whether the people arriving have the profile, timing and means to buy.
- Response speed
- How long between a contact arriving and someone answering.
- Sales progression
- How much of what comes in advances between sales funnel stages.
- Channel efficiency
- What each channel delivers relative to what it consumes.
We publish no benchmark figures for any of these. They vary with sector, ticket size, cycle, region and maturity — and an average quoted out of context leads to a wrong decision that looks well-founded.
Process
How the work happens
From the decision that has to be taken to the routine that sustains it.
Decisions and diagnosis
Mapping the decisions the company needs to take, and what is measured today, with which tool and at what level of reliability.
Measurement plan
A written definition for each indicator, the sources, the attribution windows and the success criteria.
Implementation
Events, conversions, source organisation and dashboards built inside the tools in use.
Reading
Periodic analysis separating what the data supports from what it merely suggests, with a prioritised recommendation.
Routine
A follow-up session with the people who decide, where each recommendation becomes a recorded decision — or is dismissed with a reason.
What depends on you
- Access to measurement tools, media accounts and the website.
- Commercial data: without knowing what became a sale, marketing can only be judged by platform metrics.
- Feedback from the sales team on the quality of what arrives.
- The presence of decision-makers in the reading routine. Analysis without a decision-maker becomes a filed report.
When it fits
When this area is the right step
Intelligence addresses uncertainty in decisions, not a shortage of demand.
- When there is substantial investment and no consolidated reading of the return.
- When platforms report different numbers and nobody knows which to follow.
- When you have to decide where to increase and where to cut budget.
- When marketing and sales argue about lead quality with no shared criteria.
- When waste is suspected and there is no way to locate it.
- When the company is about to scale and needs to know exactly what is being scaled.
Limits
When intelligence alone does not solve it
Measuring well shows the problem clearly. It does not fix it.
- When the volume of data is too small to tell signal from noise.
- When there is no sales record that connects marketing to revenue.
- When nobody in the company has the authority to act on the recommendation.
- When the expectation is absolute precision — which no marketing measurement delivers.
- When the real need is data engineering rather than reading. In that case, the assessment is made together with Centriu.
The analysis says where the problem is. Fixing it is usually the work of another area — media, website, content or the sales process.
How we do not work
What we do not do in this area
Common practices in the data market that we do not adopt.
- We do not build dashboards with sample data to illustrate a proposal.
- We do not present platform metrics as though they were revenue.
- We do not publish aggregate figures without source, period and definition.
- We do not hand over a dashboard with nobody responsible for reading it on a routine.
- We do not sell complex technology integration as the core service.
Where a visual example appears on this page or in a proposal, it is a conceptual structure with no data — never a client result presented as a benchmark.
Questions
Frequently asked questions about marketing intelligence
Why does each platform show a different number?
Because each uses its own attribution window, its own conversion criteria and its own ability to identify the person. Adding up different dashboards produces a number that does not exist. The work is to define one reference and use the others as diagnosis.
Do I need to buy a tool?
In most cases, no. We work inside what the company already uses. Where something essential is missing, we point to what fits the scope — but the choice and the contract are the company’s, and we do not resell platforms.
Do you deliver dashboards?
We do, when one serves a recurring decision. But the dashboard is a means, not an end: what defines the deliverable is the periodic reading and the recommendation. A dashboard with no reading routine stops being opened within weeks.
What if my operation measures almost nothing today?
That is the most common situation, and it is where the work begins. Implementing well-defined basic measurement usually pays back more in the first cycle than any sophisticated analysis of data that does not exist.
What if I need integration between systems?
Where the need involves custom connectors, data engineering or development, it is assessed together with Centriu. Arte com Pimenta does not develop software, and does not take on that scope without that validation.
Can I hire intelligence on its own?
You can, and sometimes it is the right choice — particularly when other agencies or an internal team already execute and what is missing is an independent reading. What has to exist is access to the data and someone with the authority to act on the recommendation.
Before asking for another report, it helps to define which decisions it has to support.
The measurement diagnosis starts from the decisions at stake, assesses what is already measured, and points out what has to exist for the reading to be reliable.
No proposal is sent before the diagnosis conversation.