Dominik Hergeth
← All casework

Case study · Fernhollow Analytics

The margin drop that wasn’t a problem

Company
B2B SaaS analytics platform with SMB and Enterprise customers, preparing a Series A
Data
15 months of monthly P&L (Jan 2024 – Mar 2025): revenue and new customers by segment, cost of goods sold (COGS) in four categories
Methods
Excel, gross margin decomposition, regression, payback analysis

Fernhollow Analytics is a fictional company. I designed the dataset and the problem myself, then worked through it the way the company’s finance lead would, ending in a CFO-style memo.

The question

“Our gross margin has fallen from about 80% to about 70% over the past 15 months. Our board is worried this signals a cost-efficiency or operating-leverage problem, and it’s going to be a hard question in Series A diligence. Is something breaking in how we run the business, or is this something else?”

The founder

Bottom line

Nothing is breaking. The margin decline comes from a shift in customer mix toward Enterprise, not from a cost-efficiency problem. Enterprise customers cost far more to onboard, but they earn that cost back in about two months.

Gross margin will probably keep falling while Enterprise grows as a share of the business. That is expected and not a cause for alarm, as long as the timing of cash is managed.

Gross margin
80.6% → 70.3%
Onboarding, % of revenue
5.7% → 19.9%
New Enterprise customers / mo
2 → 32
Enterprise onboarding payback
~2 months

How I got there

  1. Confirm the pattern and size it

    Gross margin fell from 80.6% in January 2024 to 70.3% in March 2025, about ten points. The decline was spread evenly across the fifteen months. There was no single bad month, so this was a trend, not a one-off shock.

    Gross margin% of revenue, monthly
    Show data

  2. Go through the costs one line at a time

    COGS has four categories. Three of them grew in line with revenue or more slowly. Hosting fell from 6.9% to 4.3% of revenue, customer support from 3.9% to 2.6%, and payment processing stayed at 2.9%. I ruled all three out.

    Implementation and onboarding went from 5.7% of revenue to 19.9%. That one line explains the whole decline.

    Cost of goods sold, split% of revenue, monthly
    Show data (all four categories)

  3. Test the obvious explanation instead of assuming it

    The easy conclusion is that the company has become worse at onboarding, meaning each customer now costs more to set up. Instead of assuming that, I ran a regression of monthly onboarding cost against the number of new customers in each segment. This separates two possible causes: each onboarding getting more expensive, or the company onboarding more of the expensive kind of customer.

    One fixed cost per customer type explains 99.9% of the month-to-month movement: about $252 per new SMB customer and $6,104 per new Enterprise customer. The cost of onboarding one customer had not changed at all.

    Onboarding cost, explained by segmentFitted cost per segment vs. actual cost, $ per month
    Show data

    Turning pointThe fitted bars match the actual cost dots almost exactly, while the Enterprise part of each bar grows. The margin trend on its own can’t show this. You only see it once the cost is split by segment.

  4. Find what actually changed: the mix

    New Enterprise customers rose from 2 a month to 32 a month, while new SMB customers stayed at around 30 a month. Onboarding one Enterprise customer costs about 24 times as much as onboarding one SMB customer. So a shift toward Enterprise pulls down the blended gross margin even though the unit economics haven’t changed.

  5. Check the other side before concluding

    A mix shift is only good news if the expensive customers pay for themselves. Average Enterprise revenue grew from about $2,850 to $2,990 per customer per month, so the $6,104 onboarding cost is earned back in about two months. The unit economics are strong. The margin line looks worse, but the business is healthier.

Why it’s easy to miss

A falling blended margin looks exactly like an efficiency problem. The two only separate once you split costs by segment and hold the cost per customer constant. If you stop at the headline, you end up fixing something that isn’t broken, and possibly slowing down the Enterprise growth that is driving the business.

The memo

What’s working

  • Every other COGS line is growing in line with revenue or more slowly.
  • Enterprise onboarding pays for itself in about two months of revenue.

What’s breaking

  • Nothing structural. The margin decline is a side effect of healthy Enterprise growth.

Forward implication

  • Expect gross margin to keep declining while Enterprise’s share of new customers rises. The board and Series A investors should hear this explanation before they ask.

Recommended actions

  • Build a rolling cash forecast that models the gap between paying for onboarding up front and collecting Enterprise revenue afterwards. As Enterprise scales, that timing gap becomes a liquidity question.
  • Set a monitoring threshold on onboarding cost as a share of revenue, so any real rise in the cost per customer is caught early.

What we don’t know yet

  • Gross margin by segment: hosting and support costs aren’t split between SMB and Enterprise.
  • Enterprise retention and lifetime value.
  • Whether the shift toward Enterprise continues or levels off.