Why is our gross margin falling?

Gross margin was 61% four quarters ago. It is 57% now. Revenue is up 34% over the same period, nobody has changed the pricing, and the cost of delivery per unit looks roughly the same as it always did.

Four points does not sound like much. On ₹12 crore of revenue it is ₹48 lakh a year, which is a senior hire, or most of a bad quarter, or the difference between a company that funds itself and one that raises again.

It is also the single most common thing a founder cannot explain from their own accounts, because a margin is a ratio and a ratio can move without any of its components doing anything obviously wrong.

Three causes, and only three

Whatever the specific story turns out to be, a change in blended gross margin decomposes into exactly three things. Getting the decomposition right before investigating saves a great deal of wasted effort.

1. 01Mix. The margin on each thing you sell is unchanged, and you are selling a different combination of them. Nothing is wrong at the line level and the blend has moved.
2. 02Price. What you actually realise per unit has fallen, through discounting, a channel that takes a cut, or a list price that has not moved while everything else has.
3. 03Cost. What it costs to deliver a unit has risen, through input prices, delivery inefficiency, or costs that were always there and were not being counted.

They require different responses. A mix problem is a sales and marketing decision. A price problem is a commercial decision. A cost problem is an operations decision. Diagnosing the wrong one is how a company puts prices up to solve a mix problem, loses the customers it most wanted, and makes the ratio worse.

Before asking why the margin fell, ask which of mix, price and cost moved. Two thirds of the investigation is usually unnecessary once you know.

The decomposition, in an afternoon

This does not need a system. It needs a revenue export by line item for two periods and a couple of hours.

1. 01Pick the two periods you want to compare. A quarter against the same quarter last year is usually cleaner than sequential quarters, because seasonality does not confuse it.
2. 02Split revenue into the segments that plausibly differ: product, plan, channel, customer size, location. Three to six groups, no more.
3. 03For each segment, in each period: revenue, volume, and cost of delivery. Calculate margin percentage per segment.
4. 04Now hold two of the three still at a time. Recalculate the blended margin using this period's mix with last period's segment margins. The difference from last period's blended margin is the mix effect.
5. 05Then recalculate using last period's mix with this period's segment margins. That difference is the rate effect, meaning price and cost together.
6. 06Finally, inside the rate effect, split price from cost: revenue per unit against cost per unit, by segment.

The two effects will not sum exactly to the total change, because there is always a small interaction term where mix and rate moved together. Ignore it if it is small, and if it is large, your segments are too broad.

A worked example

Illustrative figures for a fictional company with three segments. Blended margin fell from 60.4% to 56.5%, a decline of 3.9 points.

• ·Enterprise: 68% margin last year, 67% now. Share of revenue fell from 42% to 31%.
• ·Mid-market: 59% margin last year, 58% now. Share held at about 38%.
• ·Self-serve: 47% margin last year, 44% now. Share rose from 20% to 31%.

Every segment's margin fell by one to three points, which is a real but modest cost and price story. The blended figure fell almost four. Most of the gap is that the highest-margin segment shrank as a share of revenue while the lowest-margin one grew by half.

Run the decomposition and 2.3 of the 3.9 points are mix, 1.4 are rate, and the remaining 0.2 is the interaction between them. Which changes the entire response. There is a real cost and price story worth an hour, and the larger event is that the company's growth came from the segment it earns least on, which nobody decided and everybody contributed to.

That finding is almost always the same underneath: marketing optimised for cost per sign-up, self-serve was the cheapest to acquire, and the budget followed. Every individual decision was defensible and the aggregate was a four-point margin decline.

When it is genuinely mix

The fix is not to stop selling the low-margin segment. It is to make the trade deliberately.

• ·Check whether the cheap segment actually makes money once acquisition cost and cost to serve are both counted properly. Frequently it does not, and the growth is being subsidised by the segment that does.
• ·Look at where acquisition spend goes, by segment, against contribution rather than against sign-ups. This is usually the single highest-return change available and it costs nothing.
• ·Consider whether the low-margin segment is a funnel into the high-margin one or a destination. If customers genuinely graduate, thin early margin is an investment. If they do not, it is a leak.
• ·And decide whether to reprice it, restrict what it includes, or accept it with the economics understood.

What not to do is raise prices across the board. That applies a fix to the segments that were not the problem and leaves the mix unchanged.

When it is price

Realised price falls for four reasons, and they are worth separating because only one of them is a pricing decision.

• ·Discounting that grew. A sales team granting more, or larger, discounts than it did a year ago, usually without anyone tracking the gap between list and realised price. Measure that gap monthly and it stops widening.
• ·Channel shift. The same product sold through a route that takes a commission. An aggregator, a marketplace, a reseller. The customer pays the same and you receive less, and the margin change is entirely structural.
• ·Contracts that never moved. Retainers and annual agreements priced two years ago, with scope that has grown since. Extremely common, invisible without comparing delivered value against fee, and the fix is a renewal conversation rather than a price rise.
• ·List price standing still while costs rose. Not a fall in price so much as a failure to move one, and the effect on margin is identical.

When it is cost

The most investigated and often the smallest of the three. Worth checking in this order.

• ·Input prices. Cloud, model or API charges, ingredients, materials, subcontractors. Straightforward to verify and the first place everyone looks.
• ·Efficiency. More hours, more wastage or more rework per unit delivered. In a services business this is utilisation and realisation; in a kitchen it is wastage; in software it is usage per account.
• ·Costs that migrated into cost of sales without anyone noticing. A support team that grew with the customer base, a delivery role that was reclassified, payment gateway fees rising with volume.
• ·And costs that were always there and were not being counted. This one is not a real decline at all: the margin did not fall, the measurement improved. It matters because the response is to correct the history rather than to fix anything.

The ones that look like margin problems and are not

Three situations where the ratio moves for reasons that have nothing to do with margin.

The first is revenue recognition. A business moving from upfront billing to monthly, or recognising revenue on delivery where it previously recognised on invoice, changes the numerator without changing the economics. Check whether the policy changed before investigating anything else.

The second is a growing business with a lag. Costs that scale with volume arrive immediately while the revenue from new customers ramps over months. A fast-growing company can show a declining margin purely because the ratio contains this quarter's costs against a revenue base that has not caught up. This corrects itself and it is worth knowing that is what you are looking at.

The third is one large deal. A single unusual contract at atypical margin can move a blended figure by a point or two on its own. Recalculate with the largest customer removed and see whether the trend survives. Frequently it does not.

Doing it when you cannot split the cost

The decomposition assumes you know cost of delivery by segment. Plenty of companies do not, and the honest response is to estimate rather than to abandon the exercise.

Take the costs that clearly belong to one segment and assign them. For the rest, pick one allocation rule per cost and write it down: support cost by ticket volume, delivery cost by hours, payment fees by transaction count, infrastructure by usage where it is measured and by revenue where it is not.

The rules will be approximately right and that is enough, because you are measuring a change between two periods rather than an absolute. As long as the same rule is applied to both periods, an imperfect allocation still shows you which direction moved. A rule that is wrong in a consistent way cancels out of the comparison.

What does not work is changing the allocation between the two periods, which manufactures a margin movement that never happened. If the method has to change, restate the earlier period on the new basis before comparing anything.

A first pass on estimates, done this quarter, is worth considerably more than a precise version next year, because the mix keeps moving in the meantime.

What to do about a four-point decline

1. 01Decompose it before investigating. Mix, price, cost. An afternoon, and it usually eliminates two of the three.
2. 02Check for classification changes and revenue recognition changes between the periods.
3. 03Recalculate without the largest customer, to see whether one deal is carrying the trend.
4. 04Then investigate whichever of the three actually moved, and only that one.
5. 05Put segment margin in the monthly pack, permanently, so the next four-point move is visible in the quarter it starts rather than a year later.
6. 06And put acquisition spend by segment next to it, because the mix version of this problem is created there.

The last two are what stop it recurring. A blended gross margin reported alone is a number that can fall for four quarters before anyone can say why, and by then the mix has moved so far that reversing it is a strategy rather than a correction.

The blended margins piece on this site covers why an average hides a loss-making segment in the first place, and the unit economics work behind both is what turns segment margin from an annual exercise into a monthly line.

About Simplify

Simplify is a finance clarity and investment readiness practice working with founders across India, built on six years inside startups. We write about the questions founders bring before a decision, not after it.