What is Reach?

A campaign delivers 900,000 impressions at an average frequency of 3.6. Reach is 900,000 ÷ 3.6 = 250,000 people. Doubling the budget on the same audience will not double reach: the extra impressions largely land on people already reached, so frequency rises toward 6 or 7 while reach grows by far less than half again. Adding people requires a larger audience, not a larger budget.

Reach is the number of distinct people who saw an ad at least once during a period. Impressions count deliveries; reach counts the audience those deliveries landed on.

The relationship between the two is frequency, and the three numbers are one identity:

Reach = impressions ÷ average frequency

Any two of them determine the third, which means a report showing only one is hiding the other two. The most common version of this omission is a campaign summary leading with a large impression figure and no indication of whether it reached a hundred thousand people or ten thousand.

Unlike impressions, which are counted, reach is modelled. The platform has to decide which deliveries went to the same person, and that decision is an inference.

Where people are logged in across devices, the inference is good: the same account on a phone, a tablet and a desktop browser can be resolved to one person with high confidence. Where they are not, the platform is working from device signals and behavioural patterns, and the estimate degrades. Households sharing a television or a tablet degrade it further, since one device may represent several people or one person may appear as several devices.

None of this makes the number useless. It makes it a modelled estimate with an error band, precise enough for planning and not precise enough to be treated as a headcount. Reach figures from two different platforms were produced by two different models, which is one of the reasons they cannot simply be added.

The most consequential property of reach is that it saturates, and impressions do not.

An audience definition contains a finite number of people. Early in a flight, most impressions land on people who have not seen the ad, so reach grows almost as fast as impressions. As the campaign progresses, a rising share of impressions land on people already reached. Reach growth slows, then flattens; frequency keeps climbing.

Past that point, additional budget buys repetition exclusively. The campaign report will show impressions rising, spend rising, frequency rising, and reach essentially static — and a team measuring only spend and impressions will not notice.

This is why a reach shortfall is almost never a bidding problem. The fixes are structural: broaden the targeting, add placements, add a platform, or accept the ceiling and plan the budget around it. Calculating the addressable audience before the campaign runs, and comparing it to the reach goal, prevents most of these situations entirely.

Within an audience, the people easiest to reach are reached first: heavy users of the platform, people with high session frequency, people whose profiles match the targeting most strongly. They are cheap because they generate a great deal of inventory.

The people who remain are lighter users who generate fewer opportunities, and reaching them requires either winning more auctions or waiting longer. The cost of each additional percentage point of the audience rises steadily.

The practical consequence is that incremental reach is expensive in a way that the average cost per thousand impressions completely conceals. A campaign reporting a stable impression price can be paying steeply more for each new person it adds, and the only way to see it is to plot reach against spend over the flight rather than reading the averages at the end.

Reach cannot be summed. The same person uses several platforms, so adding platform-reported reach figures counts them once per platform.

The overlap between major platforms is substantial, and it is largest exactly where an advertiser is most likely to want a total — a broad consumer campaign running across several feeds. Summed figures in that situation can overstate true reach considerably.

Resolving it properly requires a panel, a measurement partner or a modelled overlap assumption. Where none is available, the honest presentation is a range with the summed figure as the upper bound and the largest single platform's reach as the lower one.

Most platforms offer a delivery mode that maximises distinct people rather than impressions or outcomes, usually with a frequency control attached.

It does what it says: it suppresses repetition in order to spread the same budget across more people. That is the correct setting for a launch, a category-entry campaign or an announcement where the objective is that as many people as possible hear something once.

It is the wrong setting for most other things. Several exposures are frequently what makes a message land, and a mode that actively prevents the second one will underperform for any campaign relying on repetition to build recognition or overcome hesitation.

The failure mode is choosing it by default because more reach sounds better than less. It is a trade, and the right side of the trade depends on whether the message works on first contact.

Plot reach and frequency together across the flight rather than reading end-of-campaign averages. The shape says what the averages cannot: where reach flattened, where frequency began climbing, and therefore where the campaign stopped adding people and started repeating itself.

For a brand or launch campaign, that inflection is the moment to change something — audience, creative, placement — rather than a moment to add budget.

For a direct-response campaign it may not matter at all, which is the final point: reach is not a universal virtue. It is the right metric for a specific class of objective, and for the rest it is context rather than a target.

Reach counts people and impressions count deliveries, and the distinction decides which campaigns can be judged by which number. A direct-response campaign cares little about reach: it wants the conversions, and if they come from showing the same warm audience six ads then the six ads were worth buying. A launch, a category-entry campaign or a brand campaign cares about almost nothing else, because being seen twice by the same person is not the same as being seen once by two. The asymmetry that matters in practice is that impressions can always be bought with more money and reach frequently cannot: once the addressable audience is saturated, additional budget buys frequency. That ceiling is a property of the audience definition rather than of the campaign, which is why a reach problem is almost always solved in targeting rather than in bidding.

• Treat a reach ceiling as a targeting problem, not a budget problem. Once the audience is saturated, more money buys frequency and nothing else.
• Check the audience size against the reach goal before the campaign runs. A plan that needs to reach more people than the targeting contains cannot succeed at any budget.
• Deduplicate across platforms before quoting a total. Summed platform reach counts the same person several times, and the overlap between major platforms is substantial.
• Watch reach and frequency together over the flight. Reach curves flatten while frequency keeps climbing, and the point where it flattens is where the campaign stopped adding people.
• Use reach-optimised delivery deliberately rather than by default. It suppresses repetition, which is right for a launch and wrong for a campaign that needs several exposures to work.