Treat frequency as an audience experience
Traditional television planning often accepted repeated exposure as an unavoidable by-product of buying schedules. The connected-TV paper describes the shift from uncapped frequency toward managed frequency and creative sequencing. That shift matters because every additional impression has two possible effects: it can reinforce memory, or it can waste budget and irritate a household already saturated with the same message.
Begin by defining the campaign's communication task. A launch may need enough exposure to establish a new idea; a reminder campaign may need lighter contact; a complex product may need several different messages. There is no universal cap that fits every goal, audience, creative length, or flight. The rule should follow the learning required and the tolerance for repetition, not a platform default.
Read the distribution behind average frequency
Average frequency divides impressions by deduplicated reach, but it conceals uneven delivery. Two plans with the same average can be very different: one may spread exposures relatively evenly, while another reaches many households once and repeatedly serves a small group. Approval and optimization should therefore use frequency buckets, including unreached eligible audience, light exposure, intended exposure, and high exposure.
Examine the distribution by publisher, audience segment, device group, geography, and time period. A weekly view can show whether overexposure emerged late in the flight or was concentrated from the start. Pair frequency with incremental reach: as new household reach slows and high-frequency buckets grow, the next impression is increasingly likely to repeat rather than extend the audience.
Set caps at the level identity can support
A cap is only as reliable as the identifier enforcing it. Within one logged-in publisher, exposure may be recognized consistently. Across apps, devices, supply platforms, and linear television, the campaign may depend on a household graph or modeled match with gaps. Document whether each cap applies to a device, account, household, publisher, buying platform, or broader resolved audience. Never label a platform-level cap as universal.
Use nested rules where possible: a campaign cap for the resolved audience, publisher caps to prevent local concentration, and creative caps to limit replay of one execution. When identity is uncertain, choose a conservative cross-platform assumption and use time-based rotation as a fallback. A slightly lower local cap can create room for exposure that the system cannot see elsewhere.
Build a sequence that still works out of order
The paper illustrates sequencing as message A, then B, then a conversion-oriented ad. In practice, every unit should be understandable on its own because viewers can enter late, skip a publisher, share a household, or lose identity continuity. Design an opening asset that establishes the brand and problem, a middle asset that supplies proof or differentiation, and a later asset that makes the next step clear.
Create explicit transition rules. A verified exposure can move a household forward; a completed view may be a stronger signal than a started impression; a conversion should trigger suppression or a customer-specific message. Also set a maximum dwell time so viewers do not remain trapped at one step. When cross-publisher order cannot be enforced, rotate the campaign's emphasis by phase while keeping all assets coherent.
Operate with a frequency and sequence checklist
Before launch, confirm the audience unit, identity scope, cap window, publisher overrides, creative order, transition signals, suppression logic, and reporting buckets. Test that naming conventions and event timestamps survive across the buying and measurement systems. Assign one owner to reconcile platform reports and flag where summed publisher reach or frequency cannot be treated as deduplicated.
During the campaign, review both reach expansion and high-frequency growth. If saturation rises, rotate creative, broaden qualified supply, relax a nonessential audience constraint, extend the flight, or return budget according to pre-approved rules. After the campaign, compare planned and observed distributions and document where identity failed. Those findings improve the next forecast more than an isolated average frequency figure.
| Control | Question to answer | Fallback when identity is incomplete |
|---|---|---|
| Audience cap | Which person, household, account, or device is capped? | Use conservative publisher and device limits |
| Sequence trigger | What verified event moves a viewer to the next message? | Rotate creative emphasis by campaign phase |
| Suppression | Which outcome stops or changes messaging? | Apply suppression wherever deterministic signals exist |
| Reporting | Can exposure be deduplicated across every source? | Show source-level data and label modeled totals |
Decision implication
Frequency capping and sequencing turn CTV from repeated spots into a managed communication journey. The system will not be perfectly cross-platform in every market, so good planning makes the identity boundary explicit and supplies fallbacks. Control what can be controlled, report what cannot be deduplicated, and use the exposure distribution to guide action.
Download the full Connected TV Advertising whitepaper for the measurement transformation and unified-video vision behind this playbook.
Use a PyxiVisio decision-intelligence model to connect assumptions, delivery, economics, risk and approval conditions.