What This Chart Shows
This chart shows where the next dollar (or next unit of effort) should go for each channel: into Magnitude or into Frequency. For each lever you see two bars side by side: marginal return from magnitude (reaching / expanding the audience at your current frequency) and marginal return from frequency (more impressions to the same users).
When costs are available, the chart shows mROAS (marginal Return On Ad Spend) for each path; when costs are not available, it shows marginal lift decomposed the same way (what share of lift comes from reach vs frequency).
A short recommendation above each pair tells you whether to expand magnitude, increase frequency, treat the channel as saturated, or either. So you can see, per channel, whether to invest in audience expansion or in frequency, and how strong that signal is.
Key Questions This Chart Helps Answer
- Should we invest the next dollar in reaching more people or in showing more impressions to the same people?
- Which channels are magnitude-limited vs frequency-limited?
- How does marginal return from magnitude compare to marginal return from frequency for each channel?
- Where is the channel saturated (both magnitude and frequency returns low)?
Axes, Metrics, and Units
Element | Description |
|---|---|
| X-axis | Channel (lever). One category per lever; labels truncated, full name in hover. Up to 15 levers, ordered by max(magnitude, frequency) metric. |
| Y-axis (primary) | Costs mode: Marginal ROAS (mROAS). Ratio of incremental lift to incremental spend; dimensionless. No-costs mode: Marginal Lift. Same units as the modeled KPI lift (e.g. conversions, revenue). Both are modeled. |
| Green bars | Magnitude. In costs mode: mROAS from +1% magnitude (reach) at current frequency. In no-costs: marginal lift attributed to magnitude (kernel ratio × lift). |
| Amber bars | Frequency. In costs mode: mROAS from +1% frequency at current reach. In no-costs: marginal lift attributed to frequency. |
| Recommendation text | Per lever, above the bars: "→ Expand Reach", "→ Increase Freq", "⚠ Saturated", "⚠ No signal", or "→ Either". Based on relative size of the two bars (and in costs mode, both < 1 → Saturated). |
| Breakeven (costs mode only) | Either a dashed horizontal line at 1.0 (mROAS = 1 breakeven) or diamond markers on a secondary Y-axis showing break-even in days. mROAS > 1 means return above breakeven. |
| Legend / footnote | Explains green = return from reaching new users, orange = return from more impressions to same users. |
Values are modeled. They are not raw spend or raw conversions.
Control Options Reference
Control | Meaning |
|---|---|
| KPI classification / hierarchy | Chooses which KPI/target block to use when loading R&F and attribution data (e.g. which outcome for mROAS or lift). |
| Event categories | Filter by specific event category (include/exclude). |
| Funnel stages | Filter by specific funnel stage (include/exclude). |
There is no grouping mode; the chart always shows one row per lever (channel/event).
How to Interpret the Results
- Green taller than amber: Marginal return from magnitude is higher → prioritize expanding reach (audience, placements) for that channel.
- Amber taller than green: Marginal return from frequency is higher → consider increasing frequency (often when current frequency is low).
- Both low (costs mode): If both bars are below 1.0 → saturated; the next $1 doesn't pay back; cap or reallocate.
- Recommendation: Use "→ Expand Magnitude," "→ Increase Freq," "⚠ Saturated," "→ Either" as the one-line action per channel; the 1.2× threshold avoids tiny differences driving a recommendation.
- Breakeven: In costs mode, mROAS > 1 or short breakeven days = next $1 is profitable; mROAS < 1 or "Never" = not. Use the right-axis breakeven days when shown for payback timing.
- No-costs mode: Bars show lift split; relative height still indicates whether magnitude or frequency has more headroom. Subtitle reminds you that a 50/50 split is used when the kernel has no mROAS signal.
Practical Applications for Marketers
Application | How to use this chart |
|---|---|
| Budget allocation | Put the next dollar into magnitude for green-dominant channels, into frequency for amber-dominant (if frequency isn't already saturated). |
| Channel tactics | "Expand Magnitude" → audience expansion, new placements, CPM tolerance; "Increase Freq" → raise caps or weight (check Optimal Frequency / Frequency Saturation first). |
| Saturation detection | Use "⚠ Saturated" and both bars below breakeven to identify channels to cap or reallocate. |
| Planning and briefing | Use recommendations and bar heights to brief teams on where to invest next. |
| With other M&F charts | Use with Optimal Frequency (targets), Magnitude vs Frequency Attribution (split), and Magnitude & Frequency Portfolio Diagnostic (portfolio summary). |
Common Mistakes & Misinterpretations
Mistake | Why it is a problem | How to avoid |
|---|---|---|
| Treating bars as total ROI or total lift | Bars are marginal (next $1 or next unit), not total. A channel can have high total and low marginal (saturated). | Use for "where to put the next dollar," not "which channel is biggest overall." |
| Ignoring costs vs no-costs mode | In no-costs, units are lift and breakeven isn't shown; in costs, mROAS and breakeven matter. | Check title and subtitle: "Marginal Lift Decomposition" vs "mROAS Decomposition" and the no-costs note. |
| Over-reading small differences | "→ Either" and the 1.2× rule mean small gaps don't force a strong recommendation. | Use recommendations as direction; don't over-precision magnitude vs frequency from tiny bar differences. |
| Comparing across different filters | Which levers appear and their order depend on event category and funnel stage. | Compare only under the same filters, or note when they change. |
| Using it without Optimal Frequency | This chart says where to invest (magnitude vs frequency); Optimal Frequency says how much frequency is right. | Use both: this for direction, Optimal Frequency for targets and over-/under-frequency. |
Caveats & Considerations
- Two modes: The chart runs in costs mode (mROAS, breakeven at 1.0, optional break-even days) or no-costs mode (marginal lift decomposition, no breakeven). Mode is determined by whether cost data is available. There is no separate "fallback" visualization; if data is missing for the chosen mode, a diagnostic figure is shown instead.
- Filter empty: If event/funnel filters leave no levers, the chart shows a filter-empty message.
- Top 15: At most 15 levers are shown, ranked by the larger of the two marginal values (reach or frequency). Use filters to focus on a subset.
- Recommendation logic: "→ Expand Reach" when reach value > 1.2× frequency; "→ Increase Freq" when frequency > 1.2× reach; "⚠ Saturated" (costs) when both mROAS < 1; "⚠ No signal" (no-costs) when both near zero; else "→ Either."
- Breakeven days: When shown, they depend on attribution and cost from the bridge; "Never" means breakeven is not reached. Use for relative comparison, not as a guarantee.