What This Chart Shows
This chart compares each channel’s current frequency to a model-derived target frequency and tells you whether to increase, reduce, or hold frequency. For each lever you see a bar = current frequency (impressions per user) and a horizontal line marker = dynamic target frequency from the model. So you can compare "where we are" vs "where the model says we should be."
A status label on each bar tells you what to do: Increase (current below target), Reduce (current above target), or Near target (within a band). The target is built from the saturation curve, conversion efficiency (magnitude vs frequency marginals), and diminishing returns. There is no time axis—it is a snapshot of current vs optimal frequency and the recommended direction.
Key Questions This Chart Helps Answer
- Should we increase or reduce frequency for each channel?
- How far is our current frequency from the model's target?
- Which channels are under-frequency vs over-frequency?
- What is the "right" frequency level per channel?
Axes, Metrics, and Units
Element | Description |
|---|---|
| X-axis | Lever or channel name. Category labels; up to 15 rows (top by effective exposure after filters). |
| Y-axis | Frequency (impressions per user). Modeled current and dynamic target; not raw impression counts. |
| Bar | Current frequency.
Green = Increase Red = Reduce Blue = Near target |
| Horizontal line marker | Dynamic target frequency. |
| Status text | Per bar: "↑ Increase (X%)", "↓ Reduce (X%)", or "● Near (X%)". |
| Annotation | Bar = Current |
Control Options Reference
Control | Meaning |
|---|---|
| KPI classification / hierarchy | Used to select which R&F block to use when the pipeline has per-target R&F data. This chart does not show an outcome/KPI group dropdown. |
| Event categories | Filter by specific event categories (include/exclude). |
| Funnel stages | Filter specific funnel stages (include/exclude). |
Note: This chart does not use grouping mode or outcome group in the chart spec.
How to Interpret the Results
- Green bar + "Increase": Current frequency is below the target band; the model suggests room to increase frequency.
- Red bar + "Reduce": Current frequency is above the target band; the model suggests reducing frequency (diminishing returns).
- Blue bar + "Near target": Current is within the per-bar band of the target; no strong increase/reduce signal.
- Bar above marker: Current > target → over-frequency (typically "Reduce").
- Bar below marker: Current < target → under-frequency (typically "Increase").
- Band width: The "Near" band is per bar (from exposure, mROAS, and slope). Tighter band = more precise "near target"; wider = more tolerance. The footnote shows median band (e.g. ±X%).
Practical Applications for Marketers
| Application | How to use this chart |
|---|---|
| Frequency planning | Use Increase / Reduce / Near to decide where to add or cut frequency by channel. |
| Cap and target setting | Use dynamic target and status to set frequency caps or targets in platform or planning tools. |
| Channel mix | Compare status across channels to prioritize "increase" vs "reduce" vs "hold." |
| Pairing with other charts | Use with Frequency Saturation (curve shape) and Reach vs Frequency Attribution (reach vs freq split) for full picture. |
| Reporting | Use "current vs dynamic target" and status to explain frequency strategy and next steps. |
Common Mistakes & Misinterpretations
| Mistake | Why it is a problem | How to avoid |
|---|---|---|
| Treating target as a fixed "optimal" number | The target is dynamic; it can change with data and model. | Use the footnote and treat target as model-based guidance, not a single universal optimum. |
| Ignoring the per-bar band | "Near target" depends on a band; the same gap can be "Increase" or "Near" for different levers. | Use the status text and band %; do not judge only by bar vs marker distance. |
| Treating "Increase" as "always add frequency" | Increase means current is below the model's target band; you still need to consider reach, creative, and strategy. | Use as one input; combine with reach/frequency attribution and business constraints. |
Caveats & Considerations
- Dynamic target non-positive: If the computed dynamic target is ≤ 0 for all levers, the chart shows “Dynamic target is non-positive for all levers.”
- No rows after filters: If every lever with a positive target is filtered out by event category or funnel stage, the chart shows “No rows with positive dynamic target after applying filters.”
- Filter-empty: If event category or funnel stage leave no levers, the chart shows the standard filter-empty state.
- Display limit: Up to 15 levers (top by effective exposure after filters). Use filters to focus.
- Model-derived: Current frequency comes from the pipeline; the dynamic target is from the kernel formula. Use for direction and planning, not as a guarantee of outcome.
- Uncertainty: The chart shows modeled current and dynamic target (point estimates); it does not show confidence intervals.