Reach is not people. Here’s what campaign numbers actually tell you.

A campaign report opens with a number in enormous type. It has six or seven digits, an upward arrow, perhaps a percentage increase. You are meant to feel reassured.

But what happened? Did someone see a message? Did a platform serve it? Did a video begin playing? Did someone seek more information? Was anything understood, remembered or changed?

Those are different questions. They produce different numbers. The trouble begins when we let the most impressive number stand in for all of them.

We regularly work on campaigns that need to reach large audiences. We also spend time reviewing their results, and one lesson keeps returning: a metric is only as useful as the claim you attach to it.

Consider this a practical guide to making the attachment honestly.

First, name the unit

Reach, impressions, views and clicks belong on the same report, but not in the same sentence without some explanation.

Reach generally describes how many accounts, users or estimated individuals a platform reports were exposed to content during a defined period. The precise definition depends on the platform, the reporting product and the campaign configuration. It is not automatically a verified census of human beings.

Impressions count occasions on which content or an advertisement was displayed or served under a platform’s rules. The same person may generate several impressions. Some advertising systems distinguish served impressions from viewable impressions, which use an additional visibility threshold.

Video views count playback events that meet a platform- and format-specific rule. A view on one service may require a different duration or action from a view on another. Even within YouTube’s own advertising reports, public views and billable TrueView views are different measures. A view is not necessarily a completed video, or even a deliberate decision to watch one.

Interactions might include reactions, comments, shares, saves and other actions. Ask which actions are included before treating the figure as engagement with the message. A reaction, a thoughtful question and an angry comment should not all be interpreted as the same kind of response simply because they sit in one total.

Link clicks tell you that a clickable element was activated. They do not automatically tell you whether the destination loaded, whether someone read it or whether an intended action was completed. For that, you need evidence from the destination and a consistent way of connecting the two systems.

None of these measures is intrinsically better. Their value depends on what the communication was supposed to accomplish.

The arithmetic that overstates an audience

Imagine a campaign reports 400,000 reached accounts on Facebook and 200,000 on Instagram.

A spreadsheet can add those figures in less than a second:

400,000 + 200,000 = 600,000.

What it cannot tell you is how many people encountered the campaign on both platforms. If, hypothetically, 150,000 of those accounts represented the same people across the two channels, the combined distinct audience would be 450,000, not 600,000.

That overlap is illustrative; the actual overlap is usually not visible in two separate platform summaries. Different reporting systems may also use estimates or distinct methods of identifying audiences. The point is not that the total is always wrong. It is that a sum of channel-level reach figures is a sum of reported figures, not proof of a deduplicated audience.

In our UNFPA campaign on women’s economic participation, the 2023 reporting includes 602,437 as a combined Facebook and Instagram reach figure. On the case study, we label it as combined platform-reported reach. It is the sum of the two platform figures, not a verified count of distinct people across both platforms: some of the people reached on Facebook may also have been reached on Instagram.

The figure has not become smaller. Its meaning has become clearer.

The same care applies when joining time periods. If someone saw a campaign in November and again in December, adding monthly reach totals may count that person twice. And if two campaigns reuse an audience, their separate reach figures cannot simply be added and described as the number of people who encountered the wider programme.

When a dashboard offers a properly deduplicated reach measure across placements within its own measurement scope, use it and state that scope. When it doesn’t, describe the aggregation for what it is.

More than a million views. Of what?

Views are unusually tempting because they sound like acts of attention.

Our By Us, For Us work with IOM reported approximately 7.3 million views across Meta and YouTube, alongside 8.52 million impressions and about 3.2 million reported reach on Meta. These are substantial distribution results. They do not describe three groups of people, and the view and reach figures do not even cover precisely the same set of platforms.

The story becomes useful when each number answers its own question:

• Impressions: how often was material displayed under the relevant reporting rules?
• Views: how many counted playback events occurred across the specified services?
• Reach: how many accounts or people did the measured platform report reaching within its scope?

Even then, distribution is not comprehension. A documentary excerpt beginning to play is evidence of exposure, not evidence that its viewer understood migration experiences differently afterward.

If you need evidence of comprehension or attitude change, design for it. That might mean carefully constructed surveys, qualitative interviews, an appropriate before-and-after study or another method suited to the campaign and audience. A platform dashboard cannot supply a causal claim it was never designed to test.

Stop calling every movement an outcome

Campaign reporting often jumps from we published to people engaged to the campaign made a difference without explaining what connects those statements.

A more useful distinction is between four layers.

1. Output

What did we actually produce and distribute? Films, interviews, explainers, posts, languages, placements, sessions or events. Output is about delivery.

2. Exposure

Did the material appear in front of an audience? Reach, impressions, counted views and frequency live here. Exposure is necessary for most campaigns, but it is not their final purpose.

3. Response

What did people do? Watch further, click, share, comment, request information, register, attend, ask a question. Even here, actions need context. A high number of reactions is not proof of agreement.

4. Outcome

Did something change that the campaign was intended to influence? Knowledge, intentions, enquiries, applications, service uptake, practices or other defined results. Outcomes require their own evidence, sometimes gathered outside social platforms.

The first three layers are often relatively easy to measure. The fourth is harder. That is not an excuse to relabel exposure as impact.

Sometimes the brief is explicitly about awareness or distribution. In that case, exposure is a legitimate performance objective. State it plainly. A communications campaign should not be judged as a labour-market programme, a policy reform or a social intervention if those outcomes were never part of its scope or measurement design.

What a useful results page looks like

We think the most helpful campaign reports answer six questions, in this order.

What were we trying to achieve? A public-information campaign and a registration drive require different indicators. Don’t begin with the dashboard. Begin with the brief.

Who were we trying to reach? Total volume matters differently when the goal is broad public awareness, reaching a specific community or supporting people who may need information in different languages.

What did we distribute, and where? Name the channels, dates, formats and languages. Without scope, totals are hard to interpret.

What exactly does each figure count? Label platform, unit, period and whether totals are deduplicated. Keep paid and organic contributions distinguishable where the source supports that distinction.

What response did we observe? Clicks, questions, registrations, conversations, completions or other actions, with appropriate limits on what they demonstrate.

What do we know, and what remains unmeasured? This is the line that often earns the most trust. A campaign can have successful distribution and still lack evidence about long-term behavioural change.

For example, a clear result might read:

602,437 combined platform-reported reach, Facebook and Instagram, 2023. This adds the individual platform figures and is not a deduplicated count of people.

That is less dramatic than writing 602,437 people changed their minds. It is also true to the evidence.

Measure the journey, not just the feed

A campaign can look strong inside an advertising platform while being almost invisible on the destination website. It is worth asking whether the two environments speak to each other.

Where appropriate, give campaign links consistent tracking parameters and make sure the destination measures the relevant action. Google’s Analytics guidance explains how source, medium and campaign parameters can associate an incoming visit with a campaign. That still doesn’t solve every attribution problem, but it gives you a much better starting point than interpreting platform clicks as completed website visits.

Consider a public-interest campaign that directs people to a guide. Useful questions might include whether the page loaded, whether visitors could access the guide, whether the language was appropriate and whether the information was usable on mobile. A click is the beginning of that sequence, not the conclusion.

If the campaign is sensitive, the right measurement may be deliberately less invasive. Privacy, safeguarding and the possibility of identifying vulnerable people should shape what you collect. Not every useful action needs to become a tracked individual record.

What we would change about the typical campaign dashboard

We would spend less space on percentage increases without meaningful baselines. A 2,000% rise from an almost inactive previous period may be mathematically correct but strategically uninformative.

We would separate platform-level measures before offering any combined figures. We would explain what was paid for, what was earned organically and what neither can prove.

We would show a small number of indicators tied to the campaign’s actual objective, not every figure an export provides.

And we would reserve a space for uncertainty. Missing data, changing platform definitions and unavailable outcome measures are part of the report, not embarrassments to hide in an appendix.

The best reports make the next decision easier. They tell the team what to repeat, what to test, what to stop assuming and what it still needs to learn.

The number is not the story

We do not argue for less measurement. We argue for measurement that respects its own limits.

A campaign can be beautifully conceived and poorly distributed. It can be widely distributed and scarcely understood. It can receive fewer views than expected but reach precisely the people who needed it. The numbers help distinguish those possibilities only when we ask the right questions of them.

Start with the objective. Name the unit. Show the evidence. Admit the limit.

Then tell the story the data can actually support.