Quick answer: The best programmatic frequency cap depends on objective, audience size, flight length and identity coverage. Start awareness near 2-3 impressions per person per week, use tighter windows for retargeting, and manage CTV at the campaign or household-aware level where possible. Then adjust against incremental reach, response and frequency distribution.
Frequency capping looks like one small platform setting. In practice, it controls 3 expensive trade-offs: how many people you reach, how often they see the message and how quickly the campaign runs out of useful impressions.
Set the cap too high and the same viewers absorb budget that could have reached someone new. Set it too low and the campaign may underdeliver before repetition has had time to work. CTV adds another complication: a television is often shared by a household, while identity signals vary by publisher, exchange and device.
The right answer is not one universal number. It is a hierarchy of caps, a starting range tied to the objective and a test that shows whether another exposure still adds value.
What is the best frequency cap for programmatic advertising?
For most broad-reach campaigns, 2-3 impressions per person per 7 days is a practical test starting point - not a universal benchmark. Prospecting can start lower, consideration often needs more repetition, and small retargeting pools usually need stricter daily controls. CTV should be monitored at the device or household-aware level available in the buying platform.
Use the table below to choose a starting hypothesis, then test it against your own frequency distribution and outcomes.
Objective or audience | Starting test range | Primary success signal | Main caveat
|
|---|---|---|---|
Broad awareness / incremental reach | 2-3 impressions per 7 days | Added reach, unique reach, CPM, brand lift | A low cap can restrict delivery in small audiences or premium deals |
Product consideration | 3-5 impressions per 7 days | Engaged visits, completed views, qualified actions | Rotate creative so repetition does not become fatigue |
High-intent prospecting | 4-6 impressions per 7 days | Conversion rate by frequency, CPA, assisted conversions | Attribution can make later exposures look more valuable than they are |
Retargeting | 1-2 impressions per day and 5-8 per 30 days | Incremental conversion rate, CPA, recency | Small pools saturate quickly; exclude converters promptly |
CTV household awareness | 2-4 impressions per household-aware identifier per 7 days | Household reach, completion rate, incremental reach | A device identifier is not always a complete household identity |
Short launch or event burst | 2-3 impressions per 1-3 days | Reach before deadline, search lift, direct response | Use only when timing matters and watch intraday clustering |
These are planning ranges from AdGeeks, not rules published by Google or the IAB. Raise or lower them only after checking audience size, flight length, media mix, creative rotation, deal type and identifier coverage. A reported average frequency of 3 can still hide a long tail of people who saw the ad 10 or 15 times.
Where should campaign, IO and line-item caps be set?
Use each level for a different job. In Display & Video 360, frequency caps can be set at campaign, insertion order and line-item level. The higher-level cap constrains the entities below it.
Campaign cap - control total brand exposure. Use it when multiple insertion orders, formats or inventory sources share the same audience and message. This is the best defense against separate teams or tactics each serving an acceptable frequency that becomes excessive in aggregate.
Insertion order cap - control a strategy or funnel stage. Use it for a defined budget, market, product or objective. An awareness IO and a retargeting IO may need different limits even when they sit in the same campaign.
Line-item cap - control tactical concentration. Use it to stop one publisher, deal, device type, audience or creative tactic from absorbing too many exposures. A line-item cap should not be the only control when users can enter several sibling line items.
A practical hierarchy might be 3 impressions per 7 days at campaign level, 2 per 7 days for a CTV IO and 1 per day on an individual CTV line item. These limits are not additive. The first eligible ceiling reached should constrain further serving.
Before tightening caps, review pacing, bids, inventory and audience eligibility using our DV360 optimization tips. Low delivery is not automatically a frequency problem.
How do cross-device and CTV household caps actually work?
Frequency control is only as complete as the identity available in the bid request and platform. DV360 states that mixed-inventory caps use modeling and available identifiers, including publisher-provided IDs, exchange-provided IDs and device advertising identifiers.
For CTV, DV360 supports capping across apps on the same device when a supported identifier is passed. That can include an IFA, publisher-provided ID, exchange-provided ID or shorter-lived session ID. If the identifier is unavailable, behavior and eligibility can change. A television-level cap therefore should not be described as perfect household deduplication.
Use these distinctions in planning:
Device-level cap: controls exposures tied to one connected TV or streaming device identifier.
Publisher-level identity: may improve recognition across a publisher's signed-in inventory, but coverage is publisher-specific.
Modeled cross-device cap: estimates shared identity when deterministic signals are incomplete. Treat it as a control with uncertainty, not a census.
Household cap: combines devices or users believed to share a household. The exact definition depends on the platform, identity partner and inventory source.
CTV is a shared screen. One household frequency of 4 may represent 4 exposures to one person, or 1 exposure to 4 different viewers. Pair frequency with completion, reach, brand-lift or conversion evidence instead of assuming each impression reached the same human.
For channel planning, see our connected TV advertising agency guide and CTV solution.
How should reach and repetition be balanced?
Start with the marginal value of the next impression. If impression 4 produces little incremental response but prevents a new household from receiving impression 1, reach should win. If a complex product needs explanation and response improves through impression 5, repetition may be doing useful work.
Do not rely on average frequency alone. Review:
Unique reach: how many distinct users or modeled users received at least one impression.
Frequency distribution: what share of reach received 1, 2, 3, 4 or more exposures.
Added reach from capping: how many additional people were reached because repeat opportunities were skipped. DV360 provides this at campaign level when an eligible campaign cap is active.
Outcome by frequency: how view-through rate, site engagement, conversions, CPA or brand-lift signals change after each exposure.
Time between exposures: 3 impressions in 30 minutes feel different from 3 impressions over a week.
Reach data in DV360 can take 72 hours to stabilize. Google also notes that added reach is not additive across daily rows because the same person can appear on more than one day. Wait for stable data and use a reporting window at least as long as the cap period.
Our DV360 metrics guide explains how reach, frequency, completion and conversion metrics fit together.
How should frequency caps change by funnel stage?
Awareness: protect scale first. Start with a moderate weekly campaign cap and measure added reach. Use broader creative rotation and avoid letting one premium deal consume the whole allowance.
Consideration: allow enough repetition for message recall, but vary the proof. Sequence creative from problem to benefit to evidence instead of repeating one spot. Measure engaged visits, qualified video completions and branded search alongside reach.
Conversion: make recency explicit. A recent cart visit may justify a short burst, while a visitor from 25 days ago should not receive the same pressure. Segment by time since action, exclude converters quickly and use a strict 30-day ceiling.
Retention or upsell: cap by customer state and product relevance. Existing customers may need fewer brand reminders but more precise messages. Separate service communications from paid prospecting wherever identity and consent allow.
Funnel labels do not override audience size. A B2B retargeting list of 4,000 matched users can saturate faster than a national consideration audience. Forecast eligible reach before launch and recheck it after every targeting or exclusion change.
How do you diagnose frequency waste or underdelivery?
Use the symptom, evidence and fix in that order.
Symptom | Check first | Likely frequency issue | Test
|
|---|---|---|---|
High average frequency, flat reach | Frequency distribution, audience size, overlap | Cap too loose or fragmented across entities | Add a campaign cap and compare added reach |
Low spend after tightening cap | Eligible reach, bid rate, inventory, deal commitments | Cap may be too restrictive | Relax one level only and hold other settings constant |
Good reach, weak response | Outcome by exposure and creative | Repetition is not adding value | Rotate message before raising the cap |
Strong response at higher frequencies | Recency, attribution window, audience intent | High-intent users may be self-selecting | Run a holdout or cap split to test causality |
CTV frequency looks unexpectedly low | Identifier coverage, publishers, device mix | Fragmented identity may undercount repetition | Compare publisher and device reports; use household-aware reporting where available |
Bursts within one hour | Timestamp or hourly delivery, pacing mode | The cap window is too broad to prevent clustering | Add tighter daily controls or change pacing |
Also check for structural overlap. Two campaigns with separate caps can reach the same audience. Different DSPs cannot automatically share one exposure counter. Publisher-direct CTV and open auction activity may report identity differently. Put a cross-platform planning ceiling in the media plan even when the platforms cannot enforce it perfectly.
For a broader setup review, use the DV360 audit checklist.
How should a frequency-cap test be designed?
Change one decision at a time. A clean test can compare a lower cap with the current cap across similar audiences, markets or randomized cells.
Write the hypothesis. Example: lowering the weekly cap from 5 to 3 will increase unique reach without materially reducing qualified visits.
Choose one primary metric. Use incremental reach for awareness, brand lift for memory or incremental conversions for performance.
Hold targeting, bid, creative and inventory stable. If those change too, the cap effect becomes unclear.
Use a full cap window. A 7-day cap needs at least a 7-day observation period, plus time for reporting to stabilize.
Read the distribution, not only the average. Look for changes in the heavy-frequency tail.
Check delivery and cost. A lower cap may improve reach but raise CPM or leave budget unspent in a small pool.
Document the decision. Record the winning cap, evidence, caveats and next review date.
Programmatic Guaranteed inventory needs special care. Google states that campaign-level caps can operate on a best-effort basis when contractual delivery must be fulfilled, while line-item and insertion-order caps are accounted for differently. Confirm the deal behavior before promising a hard ceiling.
When should reach and frequency settings be audited?
Audit before launch, after a material targeting or budget change, when a new DSP or publisher is added, and whenever reach plateaus while impressions continue to grow. Also audit when CTV delivery shifts between auction and guaranteed supply, because enforcement and identity coverage can change.
The output should show the current cap hierarchy, audience overlap, identifier limitations, frequency distribution, added reach, outcome by exposure and a clear test plan.
Ad Geeks will map campaign, IO and line-item controls, identify overlap and recommend the next measurable test.
What else do advertisers ask about frequency capping?
Is 3 impressions per week the best programmatic cap?
It is a useful awareness test starting point, not a universal optimum. Validate it against audience size, flight length, creative variety, incremental reach and outcomes by exposure.
Should frequency be capped daily or weekly?
Use a daily cap to prevent clustering and a weekly or 30-day cap to control cumulative pressure. Layer both when the platform and campaign structure allow it.
Does a campaign cap override a line-item cap in DV360?
Both apply. The line item has its own ceiling, while campaign and insertion-order caps constrain total exposure across the entities beneath them.
Can CTV frequency be capped by household?
Sometimes, but the definition and coverage vary. Many controls depend on device, publisher or exchange identifiers and may not represent every person or device in a household.
Why did spend fall after adding a frequency cap?
The cap removed repeat bid opportunities and the campaign could not find enough new eligible users. Check audience size, inventory, bids and deal constraints before loosening it.
How often should frequency caps be reviewed?
Review weekly during launch or rapid budget changes, then at least monthly for stable activity. Recheck immediately after targeting, creative, inventory or deal changes.
Which official sources support this guide?
Google Display & Video 360 Help, Manage frequency with frequency caps, accessed Oct 23, 2026. The page documents campaign, insertion-order and line-item caps, identifier-based capping, CTV behavior and guaranteed-deal limitations.
Google Display & Video 360 Help, Added Reach Measurement for campaigns, measurement method updated Apr 1, 2023; accessed Oct 23, 2026.
Google Display & Video 360 Help, Dimensions in reports, accessed Oct 23, 2026. It documents daily frequency dimensions and recommends using them to locate performance drop-off.
