The Reallocation Call: Auditing Grab's MMM Output
Objective: Given a supplied 5-channel MMM output table for Grab's ride-hailing and food delivery ad mix (contribution %, saturation status, adstock decay), apply the lesson's interpretation framework to flag the over-invested and under-invested channels.
You're the media analyst at Grab preparing the quarterly budget memo. Finance wants a reallocation recommendation, not a wall of regression output.
Read the response-curve and adstock outputs already fitted by the data science team, and translate them into a plain-English reallocation call.
Before you start
What you'll need
Free path (everything below is enough to finish)
Free, no account friction, handles a 5-channel weekly table easily
The process
2 steps
Step 01 of 02
The lesson's playbook step 6 says: fit the response curves, then solve for the spend mix that maximizes revenue at the current budget. A channel past its saturation point returns less per dollar than one still climbing its curve.
Grab's MMM output shows Google Search Ads contributing 9% marginal revenue per additional ₹1L/week up to ₹40L/week with no flattening, while Meta's marginal contribution drops from 8% to 2% once weekly spend crosses ₹18L. Grab is currently spending ₹22L/week on Meta and ₹15L/week on Search. What's the call?
Procedure
- Import mmm-output-grab.csv and freeze the header row
- Sort channels by 'marginal contribution at current spend' ascending
- Flag any channel whose current spend sits past its saturation point
Channel CurrentSpend SaturationPoint MarginalContribAtCurrentSpend Meta 22L/wk 18L/wk 2% Google Search 15L/wk 40L/wk+ 9% YouTube 9L/wk 12L/wk 6% TV 6L/wk 20L/wk 7% Podcasts 2L/wk 5L/wk 8%
Healthy
Search, TV, and Podcasts are all still climbing their curves, moving spend there returns more per rupee than Meta does today.
Unhealthy
Leaving Meta at ₹22L/week because 'it's always been the biggest channel' while it returns a quarter of what Search returns at the margin.
What this means
Saturation isn't a warning to cut a channel to zero, it's a signal that the NEXT rupee is better spent elsewhere.
So what do I do about it?
| Symptom | Action | Effort |
|---|---|---|
| A channel's marginal contribution is below 3% while others sit above 7% | Shift the next budget increment to the under-saturated channel, don't cut the saturated one to zero | 30 min |
Step 02 of 02
The lesson's worked example notes podcasts can carry a 6-week decay, a channel can look 'dead' in the week it's cut and still be driving sales for over a month after.
Grab paused podcast spend in week 40 for a cost review. Weekly revenue attributable to podcasts (per the model) was 4,200 in week 40, 3,100 in week 42, and 1,800 in week 46. Is this evidence podcasts don't work, or something else?
Procedure
- Plot podcast weekly spend and modeled contribution on the same chart
- Note the week spend drops to zero
- Measure how many weeks the contribution line takes to reach zero after that
Week Spend ModeledContribution 38 2000 4500 40 0 4200 42 0 3100 44 0 2600 46 0 1800 48 0 900
Healthy
Contribution decays gradually over ~6 weeks after spend stops, this is adstock carryover working as expected, not a sign the channel failed.
Unhealthy
Reading week-40's still-high contribution as 'podcasts don't move fast enough to matter' and cutting the channel permanently after one pause.
What this means
A channel's decay curve length tells you how long to wait before judging a pause, not whether the channel is broken.
So what do I do about it?
| Symptom | Action | Effort |
|---|---|---|
| A paused channel still shows contribution 2+ weeks later | Wait out the full adstock decay window before concluding the channel has no effect | 5 min |
Final deliverable
A one-page reallocation memo: which channels are past saturation, which still have room, and the adstock decay window for any channel under review.
See a reference example
Zomato Q3 MMM reallocation memo (excerpt) OVER-SATURATED: Google Display (spend 14L/wk, saturation point 8L/wk, marginal contribution 1%) STILL CLIMBING: YouTube (spend 5L/wk, saturation point 15L/wk, marginal contribution 8%) Recommendation: shift 4L/wk from Display to YouTube next quarter.
Success criteria
You're done when you can:
- Correctly flags every channel spending past its saturation point
- Does not misread adstock decay as channel failure