Hong Life CalculatorsUnderstanding the simple marketing mix attribution — concept and principle
When several channels run at the same time, such as search ads, social media and TV or online video, total revenue has to be credited back to them somehow before you can compare their return. This calculator does a simple what-if split: each channel's share of revenue is proportional to its spend multiplied by a weight that you choose for how strongly that channel responds. It then shows each channel's contribution and ROAS (return on ad spend). It works with any currency.
How it is calculated
- Weighted effort for each channel = spend × weight
- Total weighted effort W = search spend × search weight + social spend × social weight + TV spend × TV weight
- Channel contribution = total revenue × channel weighted effort ÷ W
- Channel ROAS = contribution ÷ spend = total revenue × weight ÷ W
The weights are relative: only their ratios matter. Doubling every weight gives the same split.
The principle
Proportional allocation assumes each channel's effect is linear: every extra unit spent produces the same response, scaled by the channel's weight. Real marketing mix modelling (MMM) estimates such response coefficients statistically from historical data, usually with regression and with adjustments for carry-over effects (adstock), diminishing returns (saturation), seasonality and price. This tool skips the estimation step and lets you test your own assumptions, which is useful for quick planning or for explaining how weights change the picture.
Worked example
- Spend: search $10,000 (weight 3), social $8,000 (weight 2), TV $15,000 (weight 1); revenue $90,000.
- Weighted effort: 30,000 + 16,000 + 15,000 = 61,000.
- Search = 90,000 × 30,000 ÷ 61,000 ≈ $44,262.30.
- Social = 90,000 × 16,000 ÷ 61,000 ≈ $23,606.56.
- TV = 90,000 × 15,000 ÷ 61,000 ≈ $22,131.15.
- ROAS: search 44,262.30 ÷ 10,000 ≈ 4.43×, social ≈ 2.95×, TV ≈ 1.48×.
How to read the result
Because ROAS is revenue × weight ÷ W, the ranking of ROAS is set entirely by the weights you enter; the tool cannot discover which channel works best. Use it to explore what-ifs: how much the split changes if TV's weight is 1.5 instead of 1. Compare ROAS with the break-even ROAS, 1 ÷ gross margin, to see which channels pay for themselves at your margin.
Common mistakes
- Treating weights from guesswork as measured results.
- Ignoring revenue that would have come without any advertising (baseline sales).
- Comparing ROAS across channels with very different time lags.
Frequently asked questions
Where do realistic weights come from?
From an MMM, from controlled experiments such as geo or holdout tests, or as a rough start from platform reports.
How is this different from last-click attribution?
Last-click gives all credit to the final touchpoint; this splits total revenue by weighted spend.
Should all revenue be split?
Better to split only incremental revenue above your baseline, so organic sales are not credited to ads.
Sources: Jin, Wang, Sun, Chan & Koehler, Bayesian Methods for Media Mix Modeling with Carryover and Shape Effects (Google, 2017)