What This Chart Shows
This chart shows how each channel’s effective response changes as average frequency increases, and where diminishing returns set in. Each line is one channel's frequency saturation curve so you see whether adding more frequency still pays off (steep curve) or if you're in diminishing returns (flat curve). The large dot on each line is that channel's current frequency. The curve shape (and where the slope flattens) shows why increasing frequency has diminishing returns and where the "sweet spot" is.
Key Questions This Chart Helps Answer
- At what frequency does each channel start to saturate?
- Is our current frequency still on the steep part or already on the flat part?
- Which channels have room to add frequency vs are already over-frequency?
- What is the approximate "optimal" frequency range?
- How does response differ across channels as frequency increases?
Axes, Metrics, and Units
Element | Description |
|---|---|
| X-axis | Average frequency (impressions per user). Grid from 0.1 to 20. |
| Y-axis | Effective response. Modeled; not raw conversions or revenue. |
| Lines | One line per channel/lever (up to 10). Color by lever. |
| Large dot | Current frequency: that lever's current average frequency and its effective response on the curve. |
| Annotation | "Large dot = current frequency · Slope change = diminishing returns onset." |
Control Options Reference
Control | Meaning |
|---|---|
| KPI classification / hierarchy | Used to resolve which R&F block to use when the pipeline provides per-target R&F data. |
| Event categories | Filter by specific event categories (include/exclude). |
| Funnel stages | Filter by specific funnel stages (include/exclude). |
Note: This chart does not use outcome group or grouping mode.
How to Interpret the Results
- Steep slope: More response per extra frequency; channel can often support more frequency before saturation.
- Flat slope: Diminishing returns; extra frequency adds little response. "Slope change" in the curve is where this sets in.
- Large dot on steep part: Current frequency is below saturation; room to grow frequency if reach/frequency strategy allows.
- Large dot on flat part: Current frequency is already in the saturated zone; consider shifting to reach or other levers.
- Compare lines: Channels that stay steeper at higher frequency have more "headroom"; flatter curves saturate earlier.
- Magnitude: Higher y = more effective response at that frequency in the model. Compare both slope and level across levers.
Practical Applications for Marketers
Application | How to use this chart |
|---|---|
| Frequency planning | Use curve shape and current-frequency dot to decide where to add or cut frequency by channel. |
| Reach vs frequency | Pair with reach–frequency attribution: flat frequency curve → emphasize reach; steep → frequency still pays. |
| Channel mix | Compare curves across channels to see who has frequency headroom vs who is saturated. |
| Creative/cap strategy | Use saturation points to set frequency caps or targets by channel. |
| Reporting | Use "diminishing returns on frequency" and "current vs optimal" to explain frequency strategy. |
Common Mistakes and Misinterpretations
Mistake | Why it is a problem | How to avoid |
|---|---|---|
| Treating "response" as raw conversions or revenue | Y-axis is modeled effective response from the PWL saturation curve, not observed conversions or revenue. | Use axis label "Effective Response"; treat as model output, not literal KPI. |
| Treating "frequency" as raw observed frequency | X is the input to the saturation model; it may be aligned with observed frequency but is the model's frequency axis. | Use axis label "Average Frequency (impressions per user)"; interpret in line with your Reach &Frequency setup. |
| Ignoring the current-frequency dot | Without it you do not know where "today" is on the curve. | Use the large dot to judge whether you are under-, at, or over-frequency. |
| Comparing across different pipelines or R&F specs | Different pipelines give different shapes. | Compare only within the same model and R&F configuration. |
Caveats and Considerations
- Filter-empty: If event category or funnel stage filters leave no levers, the chart shows the standard filter-empty state (no curves).
- No outcome/KPI group selector: This chart is not grouped by outcome; the KPI group dropdown is not shown for this chart. KPI classification/hierarchy can still affect which R&F block is used when the pipeline provides per-target blocks.
- No fallback: There is no fallback when data is missing or the contract fails. You only get the diagnostic figure until the pipeline has run and outputs are valid.
- Display limit: Up to 10 levers are shown (top by effective exposure after event/funnel filters). Use filters to focus on the channels you care about.
- Frequency range: X-axis is 0.1–20 (impressions per user). Channels with current frequency above 20 will have their current dot at the right edge; the curve still extends to 20 only.
- Model-only: Curves are from PWL-Sat parameters and R&F data produced by the pipeline; they are not raw observed response. Use for direction and relative saturation, not as a literal forecast without validation.
- Uncertainty: The chart shows modeled PWL curves (point estimates); it does not show confidence intervals.