Frequency, Reach and Saturation: Reading the Curve Before You Scale Spend
The 60-second version
Scaling spend inside a finite audience buys frequency, not reach. How to read the three curves together, and the diagnostic that tells you which wall you have hit.
- What happened, in one line
- What to do about it this week
- What you can safely ignore
You doubled the budget on the campaign that was working. Spend went up 100%. Conversions went up 31%. CPA went up 53%.
The instinct is to blame the creative, or the algorithm, or a competitor bidding up the auction. Usually it is none of those. Usually you bought the same people twice.
Reach, frequency and saturation are three views of one constraint: your addressable audience is finite, and once you have reached it, additional spend can only buy repetition. The diagnostic that tells you which wall you hit takes one chart and three columns.
The three quantities
Reach = unique people who saw the ad at least once
Impressions = total times the ad was served
Frequency = Impressions ÷ Reach
Frequency is not a setting. It is what happens to the ratio when one of the other two moves. That framing matters because it makes the diagnostic obvious: when spend rises, watch which of reach and impressions moved.
Where the Extra Spend Went
Data JourneyReach grew with spend
New people are still available at roughly the price you were paying. Frequency holds steady. This is what scaling is supposed to look like.
Frequency grew instead
The audience is exhausted. Extra budget is buying repeat impressions to people who already saw the ad and did not act.
Neither grew much
You are losing auctions rather than running out of people. CPM rose, delivery did not. A different problem with a different fix.
Those three patterns need three different responses, and telling them apart is the entire job.
The diagnostic
Show query
Never sum daily reach. Reach is a distinct count of people, so daily figures cannot be added — anyone who saw the ad on Monday and Wednesday is counted twice. Take the platform's own weekly or monthly deduplicated reach figure. This is the most common error in home-built frequency reporting, and it makes frequency look artificially low exactly when it is climbing.
Reading the saturation curve
The relationship between spend and incremental reach is not linear. It bends, and the bend is the thing you are trying to locate.
One campaign, four weeks of scaling
The number that makes this legible is cost per person reached. Unlike CPM, it is not diluted by frequency:
- Week 1: ₹0.59 per person reached
- Week 4: ₹1.06 per person reached
Almost double, for the same people. CPM barely moved across those four weeks, which is why CPM alone would not have caught it.
What frequency actually does to performance
Frequency is not simply bad. The relationship is a curve with a useful middle.
| Frequency (weekly) | Typical effect |
|---|---|
| Under 1.5 | Under-delivered. Most of your reach saw the ad once, which is rarely enough for anything considered. |
| 1.5 – 3 | The productive band for most prospecting. Enough repetition to register, not enough to annoy. |
| 3 – 5 | Diminishing. Incremental conversions per impression fall noticeably; still defensible for retargeting or a short, high-intent window. |
| Above 5 | Usually value-destroying on prospecting. Negative feedback rises, relevance scores fall, and CPMs rise as a result — a compounding penalty. |
Averages hide the tail. A campaign at frequency 4.0 rarely means everyone saw it four times. It usually means most people saw it once or twice and a small group saw it fifteen or twenty times. That heavy tail is where negative feedback and ad fatigue concentrate. If your platform exposes a frequency distribution rather than just an average, use it — the average is the least informative summary of a heavily skewed distribution.
The three responses
Match the Fix to the Wall
Process FlowSaturated — expand the audience, not the budget
Broaden targeting, add geographies, add a lookalike at a wider percentage, or add a placement you have excluded. More budget into the same audience only buys frequency.
Saturated and the audience is genuinely maxed — change the creative
A new creative resets fatigue on an audience that has seen the current one too often. This buys real incremental reach-equivalent performance without a wider audience, and it is the cheaper intervention.
Auction-limited — raise the bid or reconsider the value
CPM up with reach and frequency flat means you are losing auctions. Either the impression is worth more than you are bidding, or it is not and you should accept the lower delivery.
Still scaling — keep going, and watch the curve
If reach is still growing roughly proportionally to spend, you have not found the wall yet. Increase in steps and re-check, rather than doubling and hoping.
Scaling in steps, and why
The reason to increase spend in 20–30% increments rather than doubling is that the diagnostic above needs a clean before-and-after. Doubling the budget changes delivery so much that the algorithm re-enters a learning phase, and you cannot separate the saturation effect from the re-learning effect. Two 30% steps a week apart tell you far more than one 70% jump.
The honest caveat: frequency is correlational
Everything above is observational. Frequency rising alongside CPA rising is consistent with saturation, and it is also consistent with the algorithm exhausting the cheapest, highest-intent portion of your audience first — which would raise CPA regardless of repetition.
Both explanations are usually true simultaneously, and they have the same practical response, which is why the diagnostic is still useful. But if the decision is large — a substantial budget reallocation, or a decision about whether a channel scales at all — the only instrument that settles it is a holdout.
Show query
Compare the two period_over_period_index values. If exposed geographies grew 1.18x and holdout geographies grew 1.14x, the campaign's incremental contribution is the 4-point difference — not the ROAS the platform reported.
Frequently asked questions
What frequency should I cap at?
Rather than importing a number, find the point where your own incremental conversions per impression stop justifying the spend — that is what the weekly diagnostic locates. As a starting point, 2–3 per week is a common productive band for prospecting and higher is defensible for retargeting, but the right cap is campaign-specific and moves with creative freshness.
Does a frequency cap actually help?
It stops the waste, but it does not create reach. Capping frequency on a saturated campaign means the budget cannot be delivered at all, so it under-paces. The cap is a guardrail; the fix is a wider audience or a new creative.
Why did CPM rise when I scaled?
Two possible reasons, and they are distinguishable. Either you are bidding into more competitive inventory to find new people — in which case reach is still growing — or your relevance and engagement fell from over-exposure, and the platform is charging you more for the same inventory. Check whether reach grew: if it did not, the second explanation is the likely one.
Should I look at reach or impressions?
Reach for audience questions, impressions for cost questions. Frequency is the bridge. Most reporting shows impressions and CPM and omits reach entirely, which is precisely why saturation goes unnoticed — the omitted number is the one that would have shown it.
How does this interact with broad targeting?
Broad targeting delays saturation substantially, because the addressable pool is much larger. It also makes the frequency signal less useful as an early warning, since frequency stays low while the algorithm works through a large audience whose later portions may be far less responsive. On broad campaigns, watch cost per person reached and CPA rather than frequency.
The summary
- Reach, impressions and frequency are one identity:
frequency = impressions ÷ reach. When spend moves, which of the other two moved is the diagnostic. - Spend up + reach up = still scaling. Spend up + frequency up = saturated. Spend up + CPM up with flat reach = auction-limited. Three walls, three different fixes.
- Never sum daily reach — it is a distinct count, and summing it hides exactly the climb you are looking for.
- Cost per person reached is the clearest saturation signal, because unlike CPM it is not diluted by frequency.
- Frequency averages hide a heavy tail; a campaign at 4.0 is usually most people at 1–2 and a small group at 15+.
- Scale in 20–30% steps so the diagnostic stays readable, and settle large decisions with a geo holdout rather than a correlation.
Budget Pacing Calculator
Project whether a campaign will land under, on, or over its monthly budget, based on spend-to-date and days elapsed.
Chinmay Raibagkar
About author →Founder of DataLens AI. He helps non-technical teams read their ad and database numbers with confidence — which number to trust, what to do next, and what to ignore.