How to Tell Whether a Marketing Campaign Actually Worked

At the end of every month a business owner receives a report from their agency. Reach: three million. Engagement: a hundred and twenty thousand. Clicks: forty thousand. The figures are large, the charts point upward, and the one question the report does not answer is whether the company sold more because of the money it spent.

This is the framework we use across the group to judge a campaign, and why we reject most of what appears in a standard report.

The problem: metrics chosen because they flatter

Reach, impressions and engagement are described in reports as awareness metrics, and in theory that is a legitimate role. In practice they are reported for a different reason: they are always large and always rising. Any advertising spend increases reach automatically. Reach going up proves nothing except that you paid.

A metric that cannot go down is not a metric. It is a receipt in disguise.

Rule one: one number decides

Every campaign we run has a single primary metric, agreed before launch. Everything else is explanatory. When everything matters, nothing does, and when metrics multiply, every party finds a number to defend itself with.

The choice follows the business model:

eCommerce. Cost per acquired order against average order value and margin. The only question worth asking is whether what you pay to win an order sits below the margin it earns, with room left for operations.

Services and B2B. Cost per qualified lead — not every inbound message, but those matching criteria agreed in advance: sector, budget range, timing.

Apps and subscriptions. Cost per user who completed the core action on day one, not cost per install. An install with no use is a pure cost.

Retail with a physical location. Calls and messages that lead to a visit, with one source question asked at the point of sale.

Rule two: measurement is built before the spend, not after

The most expensive mistake we see is launching a well-funded campaign against a site that does not measure conversions at all. A month later the question "did it work?" is unanswerable, and the money spent did not even buy knowledge.

Before the first unit of currency is spent, three things must be true: conversion events are defined precisely (purchase, checkout completion, form submission, conversation started), server-side tracking is in place wherever browser restrictions swallow events, and every source is tagged consistently so channels stop claiming credit for each other's sales.

This takes days and saves months of argument. Anyone who refuses to do it and rushes to launch is selling activity rather than outcome — which is why Nsweq does not begin before tracking and the purchase path are verified.

Rule three: separate effect from coincidence

A campaign is never the only reason sales moved. Season matters. A competitor's price matters. Stock levels matter. Somebody's viral post can move demand for a week.

Two practical ways to separate them without statistical machinery:

A deliberate silence. Pause one channel for seven to ten days and watch what happens to orders. If nothing changes, that channel was harvesting demand that would have arrived anyway. It is an uncomfortable test and the most honest report you will ever read.

Geographic comparison. Run in specific governorates or cities and hold back comparable ones, then compare. Across the markets we operate in this is practical, given the spread of geography and the similarity of certain cities in buying behaviour.

Rule four: the window differs by channel

Judging SEO after a month is unfair. Judging a performance campaign after six months is waste.

Paid media is assessed two to four weeks after leaving the learning phase. Changing targeting every three days prevents the platform from learning at all, and it is the most common mistake an impatient owner makes.

Content and SEO are assessed after four to eight months, and the early indicator is not ranking position but the number of terms entering the top three pages and the movement in search impressions. We cover that in our piece on SEO after AI, and the work itself sits in Naaktob.

Brand identity is not measured by campaign at all, but by second-order effects: how quickly assets are produced once the system exists, and how consistent they are across different producers. That is covered in what you actually pay for in an identity.

Rule five: work back from break-even, not from cost per click

The number that should be pinned above your desk is not cost per click. It is the maximum allowable customer acquisition cost.

The arithmetic is simple. Take average order value, multiply by gross margin, multiply by expected purchases in a year, then subtract a share for operations. What remains is your ceiling.

A worked example: average order 1,000 units of currency, margin forty per cent, so 400. The customer buys twice a year on average, so annual value is 800. Reserve half for operations and growth and your ceiling is 400 per acquired customer. Any campaign bringing customers in below that is working even if the click looks expensive, and any campaign above it is failing even if the click looks cheap.

This single calculation ends most disputes between an owner and their agency, and it also determines how the budget should be spread across the year — the subject of our budgeting piece.

What a good report does

The monthly report we send opens with one sentence: what happened to the primary metric and why. Then no more than three explanatory figures. Then a section titled what did not work, which is mandatory: an experiment stopped, an audience that failed, a message that fell flat. A report with no failure in a full month is not a report; it is a sales presentation.

Finally, the decision: what changes next month, with an owner and a date.

The short version

Ask your agency three questions before the next report. What is the agreed primary metric? What is my maximum allowable acquisition cost? And what did not work this month? If any of the three produces hesitation, the problem is not the campaign — it is the basis on which it is being judged.

If you would like that basis built with you, tell us where you are.