Resources · Discount elasticity
How to measure price elasticity?
We have seen marketeers measure discounts against orders, revenue, and ROAS for one SKU to calculate price elasticity. On the surface all three methods look like elasticity analysis.
They are not.
The only valid methodology is to measure Orders-on-Ad-Spend (OOAS) against Discounts. Here's how the maths works out.
The setup
We tracked discounts, orders, revenue, ROAS, and OOAS for 12 weeks for bracelet SKU. All data tracked with BooleanMaths and evaluated using BooleanMaths MCP.
Approach 1 : Orders vs Discount
Logic: More discount → cheaper price → more people buy.
Result: A weak fit, R² of 0.14.
Since weekly ad spend swings from 16K to 31K, it results in orders fluctuating with it, burying any real signal under noise. Order volume changes are an effect of two forces - discount and Ad spend. And this approach only accounts for one.
Approach 2: Revenue vs Discount
Logic: More Discount → Less AOV & More Orders → So maybe we should look at Revenue instead.
Result: No correlation, R² of essentially 0.00.
It maybe tempting to conclude that "discounting doesn't help revenue because lower AOV and more Orders balance each other out" but that is not the underlying mechanism here. Revenue is also impacted by Ad-Spend and unless Ad-spends remain the same (which they did not). A flat line here means confounding, not a real insight.
Approach 3: ROAS vs Discount
Logic: Marketeers love measuring ROAS against everythin. It's the gold standard of metrics. It normalizes for spend by dividing revenue by ad spend.
Result: a strong fit, R² of 0.83, clearly positive.
This looks like the answer - but ROAS solves for one problem while introducing another. Ad-spend is adjusted for, but the discount itself becomes a part of revenue calculation.
Most importantly this approach still ignores customer psychology.
The one that works
Approach 4: OOAS vs Discount
Customer Psychology: A discount is targeted at incentivisng a hesitant customer to buy.
The only relation we need to measure is
Does a higher Discount yield more Orders?
However to measure this relationship we need to adjust for the other force that impacts Order Volume. Ad Spend.
Orders per ₹1,000 ad spend addresses all these requirements.
Result: the strongest fit of the four, R² of 0.90, across the observed 42%–52% discount range.
Orders per ₹1,000 ad spend = −29.16 + 0.872 × discount%
From orders-per-ad-rupee to revenue and margin
When you have established the linear relationship between Discount and OOAS, then discount's impact on revenue & margin is just evaluating the underlying maths.
If you assume that List Price (850) and Fulfilment Costs (COGS + Shipping + RTOs + Payments = 380) are both constant.
ROAS and Margin on Ad Spend both can be computed as quadratic functions of Discount with a Maxima followed by negative returns for higher discounts.
For the above SKU -
Revenue per ₹1,000 of ad spend keeps climbing across the tested range, toward a modeled peak near 66.7% discount.
Margin tells a different story. Every extra order also costs ₹380 to fulfill, which pulls the optimum back substantially: the margin-maximizing discount lands at 44.2% - squarely inside the observed range, so it's a real number, not a guess.
Margin peaks around ₹862 per ₹1,000 of ad spend there, then declines, turning negative past roughly 55% discount.
The takeaway
This SKU currently runs at about 46% average discount - just past the margin peak. Not a crisis, but a couple of points of margin efficiency left on the table. The broader lesson: if you are not adjusting for Ad-spend or looking at the correct metric (Orders) then you don't have an elasticity curve. You have a random scatter plot with a trend line drawn through it.
