There is a number every CFO wants but few CFOs have: the rupee return on the next rupee of trade spend. Promotion schemes get approved on instinct and last quarter's volume push. They get reviewed on the dispatch number, not the sell-through. By the time the audit comes back, the next month's plan is already approved.
This is not a failure of intelligence. The data exists. The problem is that the data sits in four different systems, the schemes overlap across products and retailers, and the lift you see in the volume report includes the bit you would have sold anyway. Unless someone is decomposing the noise, the answer of "did that scheme work" stays a matter of opinion.
The four ways trade spend leaks
When we audit promotion spend inside Indian consumer brands, the leak shows up in four buckets. The shares vary, but the buckets are always the same.
Cannibalisation is usually the largest. The buyer who was going to buy your 250 gram pack at margin instead takes the discounted 500 gram pack. You moved volume, your spreadsheet smiles, but the margin per buyer dropped.
Pull-forward is the second leak. The buyer stocks up at the discount, then the next month's volume is lower. The brand manager sees a great month, the next month looks weak, and the team blames seasonality.
Subsidy is the leak that nobody likes to talk about. A meaningful portion of the buyers taking your scheme would have bought at full price. You gave away margin to people who were already in the cart.
Inefficient targeting is the last bucket. A scheme runs in a region or at a retailer where the volume curve was going to look the same with or without it.
Add the four and you typically find 50 to 80 percent of trade spend is producing less than its face value. The question is which 50 to 80 percent in your business, which is what the rest of this post is about.
Why honest ROI measurement is hard
Three reasons, in order of how often they break the analysis.
The counterfactual problem. You cannot observe what would have happened if the scheme did not run. You can only see what did happen. To measure incrementality, you need a control group. Most brands do not have one because every retailer ran the same scheme.
Overlapping schemes. Trade marketing, channel partner schemes, retailer-led discounts, festival pricing, e-commerce flash deals, BTL activation, all running at the same time on the same SKU. The lift you see is the sum of all of them. Attributing it to any one is guesswork until you separate the time periods or the geographies.
The data lives in different systems and time grains. Sell-in lives in your ERP, sell-out lives at the retailer or in a Nielsen file, scheme spend lives in finance, modern-trade data lives in your KAM team's spreadsheets. Joining them by SKU, retailer, and week is the unglamorous prerequisite to every honest analysis.
The right way to decompose the lift
The mental model is simple. Observed lift is not the same as incremental volume, and incremental volume is not the same as profitable incremental volume. You have to decompose.
Step one is separating baseline from incremental. The baseline is what you would have sold without the scheme. You estimate it from the same period the year before, the trailing average, or a comparable retailer that did not run the scheme.
Step two is removing cannibalisation from the incremental. If the 500 gram pack went up but the 250 gram pack went down, the net incremental is smaller than the gross. The same logic applies across substitutes within your portfolio.
Step three is converting the truly incremental volume into margin and comparing it to the scheme spend. This is your ROI number. If it is less than zero, the scheme cost more than it added. If it is greater than zero, the question is whether the next rupee would have produced more elsewhere.
A five-step start, doable this quarter
If your business has never measured promotion ROI honestly, do not try to measure everything at once. Pick one scheme, one product, one retailer or region, and do the work end-to-end. The learning compounds.
- Pick a scheme that ran for at least four weeks on a specific SKU through a specific retailer. Avoid schemes that overlapped with other promotions on the same SKU.
- Pull sell-through data for that SKU and retailer for three months before and three months after. Daily or weekly grain, not monthly.
- Find a comparable retailer or region that did not run the scheme. Use that as the control to estimate the baseline.
- Subtract the cannibalisation from related SKUs. If your 500 gram went up by 100 units and your 250 gram went down by 60 units, the net incremental is 40, not 100.
- Compute the gross margin gain on the net incremental volume, divide by the scheme spend including all visible discounts and the trade marketing accruals. That is your ROI number.
The first time you do this, it will take three to four weeks. After that, the same analysis on a new scheme takes a few days.
What you find when you actually do this
Across the brands we have run this work for, the pattern is consistent. A few schemes pay back well, most break even, and a meaningful portion lose money outright.
The festive bundle paid back at 182 percent ROI because it brought in genuinely new buyers who later repeated at full price. The cashback paid back at 84 percent. The buy 3 get 1 scheme just shifted volume across SKUs. The modern trade discount lost money: the retailer used the spend to push competing brands too, the incremental did not show up.
Once you have the breakdown for the schemes that ran last quarter, the next planning meeting becomes a different conversation. The schemes that lost money do not get re-approved. The spend gets shifted to the schemes that paid back, or held in reserve. The 30 to 40 percent of trade spend that was leaking starts working.
What to do this week
Find the scheme spend file from the last quarter. Identify the three biggest schemes by spend. Pick the one where you have the cleanest data and the cleanest control. Run the five steps above for that one scheme. The number you get back is either a vindication or a wake-up call. Either way, it is more than you knew before.
If you want help running this for a specific scheme, with the data joining and the decomposition done by a senior team that has done it many times, talk to us.