Analytics & Attribution
Your Reach Number Hides How Often the Same People Saw the Ad
Updated: October 27, 2026
Average frequency can hide a concentrated pocket of overexposure. Google Ads API v26 adds unique-user reach counts at thresholds such as 2+, 3+, 4+, 5+ and 10+ impressions, so analysts can see how many people moved from first exposure into reinforcement or saturation and connect those tiers to a creative, audience or budget decision.
What the new metrics reveal
Google’s October 14, 2026 release notes added frequency-specific unique-user reach metrics. Instead of only reading total reach and an average, teams can quantify users who received at least a given number of impressions. The thresholds turn one blended metric into a distribution.
They do not, by themselves, define the “right” frequency. The useful level depends on buying cycle, format, message, audience size, seasonality and how quickly the creative changes.
Read thresholds as a cumulative ladder
| Tier | Question | Possible action |
|---|---|---|
| 1+ | How broadly did the campaign reach? | Assess scale and audience access |
| 2+ | How many received reinforcement? | Check whether message needs repetition |
| 3–5+ | Is exposure concentrating? | Compare outcome lift and fatigue signals |
| 10+ | Who is heavily exposed? | Inspect audience, exclusions and rotation |
Because the counts are cumulative, the 5+ group sits inside the 2+ group. Derive bands carefully if you need “exactly 2–4” rather than “at least 2.” Document every calculation.
Pair frequency with business response
Build a weekly view with spend, reach thresholds, qualified conversions, conversion value, brand or lead-quality indicators and creative age. Watch how the mix changes. If the 10+ audience grows while qualified outcomes flatten, the campaign may be buying repeated attention without additional value. If high-frequency users continue to convert profitably during a short launch, repetition may be supporting the decision.
Do not diagnose fatigue from frequency alone
Creative fatigue is an outcome pattern, not a universal impression count. Look for declining response, higher acquisition cost, negative feedback, lower attention or weaker post-click quality while exposure concentrates. MTC’s concept-versus-format fatigue guide can help decide whether to replace the idea, adapt the format or change delivery.
Segment before acting
- Campaign and objective.
- Creative concept and asset age.
- Placement or inventory type.
- Device and geography.
- Prospecting versus remarketing role.
- Short versus long buying cycle.
A remarketing campaign can reasonably show a different distribution from broad prospecting. One account-wide threshold can hide this difference.
Create explicit decision rules
A useful rule has a trigger, confirmation and action. Example: if the share of reached users at 10+ rises for two reporting periods, qualified conversion per reached user falls and no launch-specific explanation exists, rotate the concept or expand eligible reach. If outcomes remain healthy, document why the current exposure is acceptable.
Do not automate a budget cut from one frequency metric. Privacy thresholds, modeled values and reporting latency can affect reach reporting; use Google’s field definitions and enough volume for the comparison.
A simple reporting sequence
- Confirm the campaign and date range are comparable.
- Pull 1+, 2+, 3+, 4+, 5+ and 10+ reach where available.
- Calculate cumulative shares against 1+ reach.
- Annotate creative launches and major audience changes.
- Pair the distribution with qualified outcomes.
- Record the action and recheck after the change.
Frequently asked questions
What frequency is too high?
There is no universal number. Define the risk from audience role, buying cycle, creative, outcomes and negative signals.
Can the thresholds be added together?
No. They are cumulative: a user in 10+ is also included in lower “at least” thresholds.
What should a dashboard emphasize?
Show qualified outcomes beside the distribution and annotate creative or delivery changes. Frequency without response is incomplete.