The full 15-step worksheet run on a small cap: how to screen for a market too small to attract giants and too hard for copycats, quantify the moat in two ratios, and record the failures rather than argue them away.
1. Screen for the Goldilocks niche: too small for the giants, too complex for the copycats
The repeatable method
- Identify the largest credible competitor and ask why it has not entered. If the honest answer is "the market is not worth their time," that is a moat.
- Ask separately why a low-cost generic manufacturer has not entered. If the answer is technical or regulatory complexity, that is the second half of the moat.
- Confirm both by looking for the absence of price competition — a 70%-plus gross margin sustained for years is the evidence.
- Check the niche is still growing at least mid-single digits, so the incumbent is not defending a shrinking pond.
Here: "Rather than competing head-on with medtech giants like Medtronic or Boston Scientific, they focus solely on an interesting niche market: peripheral vascular surgery. It's a niche that's too small for the giants, but too complex for generic manufacturers." Confirmed by 71.0% gross margin and 21.3% ROIC, in an end market growing 6-7% (vascular tools) and 8-9% (minimally invasive).
Watch for
- A niche that stops being small. Success is the thing that eventually attracts the giant.
- Concentration as the cost of the strategy — the article lists it as a risk: "LeMaitre is still concentrated in one specific domain."
2. Price the switching cost by what breaks if the switch goes wrong
The repeatable method
- Work out the full cost of changing supplier — not the price difference, but the retesting, retraining and revalidation.
- Ask what the downside of a failed switch is for the person who signs it off. Where that downside is career-ending or life-threatening, no discount wins the account.
- Look for a regulatory gate (FDA, CE) that makes the alternative expensive to create, not just expensive to adopt.
- Test the conclusion against retention and pricing rather than against management's description.
Here: "Medical devices must pass through FDA and CE approvals, which can take years and can cost millions of dollars.
Switching suppliers means retesting, retraining, and risking patient safety. That's why surgeons stick with what they trust. In other words, once LeMaitre's devices are in the Operator Room, they tend to stay there." The same structure appears elsewhere in the archive as
Thermo Fisher's "massive regulatory switching costs" and
Perimeter's fire-retardant qualification.
Watch for
- Switching costs described but never evidenced. Ask for retention or repeat-purchase data.
- A change in surgical practice rather than in supplier — the article's fourth risk, and the one no switching cost protects against.
3. Quantify the moat in exactly two numbers before believing the story
The repeatable method
- Write the qualitative moat case first, then stop and demand two figures: gross margin and ROIC.
- Apply fixed thresholds — gross margin >40%, ROIC >15% — the same for every company.
- Treat a moat story that fails either test as a description, not a moat.
- Where ROE and ROIC diverge sharply, work out which is distorted and by what (cash, leverage, goodwill) before drawing a conclusion.
Here: "Let's now quantify the moat: Gross Margin: 71.0% (>40%? ✅); Return On Invested Capital: 21.3% (>15%? ✅)." Then in step 8 the divergence appears: ROE 14.9% (>20%? ❌) against ROIC 21.3% (✅), with Return on Capital Employed 11.0% and Return on Tangible Assets 10.7% alongside — the signature of a business carrying net cash and acquired intangibles.
Watch for
- ROIC flattered by a net cash position that also depresses ROE. Both are true; neither alone describes the business.
- Goodwill creeping up as the roll-up continues — here still conservative at 11.0% of assets against a <20% bar.
4. Record the failed criteria in the same table as the passes
The repeatable method
- Publish a pass/fail mark against every threshold, including the ones the company misses.
- Do not adjust the threshold to fit a company you like — carry the ❌ into the score.
- Say plainly what the failure means for a shareholder, in one sentence.
- Let the total score absorb it, so the summary number is not a curated one.
Here: three ❌ marks survive into an
8.0/10 Total Quality Score —
ROE 14.9%,
SBC 14.0% of net income (15.6% five-year average) with the flat comment "LeMaitre Vascular uses a lot of Stock-Based Compensation, which is negative to see as an investor," and the reverse DCF's
16.1% required growth marked "Realistic growth expectations? ❌". Compare
HEICO, where the same honesty produced a 7.8/10 with valuation scored 2/10 and an explicit pass.
Watch for
- A high total score that hides the location of the failures. Two of these three sit in valuation and dilution — the two that decide the answer.
- A five-year average worse than the current year (SBC: 15.6% vs 14.0%) — the problem is structural, not a one-off grant.
5. Run three valuations that can disagree, and say what the split means
The repeatable method
- Compare the forward multiple with the company's own ten-year average — a quick, admittedly crude indication.
- Build an Earnings Growth Model from disclosed inputs: EPS growth + dividend yield ± the annualised effect of a multiple you assume will fall.
- Invert with a reverse DCF: solve for the growth the current price requires, and judge that against history.
- When two agree and one dissents, treat the dissenter as the risk statement rather than discarding it.
Here: 32.5x forward against a 38.0x ten-year average ✅; Earnings Growth Model "= 13% + 1% + 0.1((25.0x – 32.5x)/32.5x) = 11.7% ✅" — note the 13% EPS assumption is a deliberate haircut on the analysts' 21.5%, and the multiple is assumed to decline to 25.0x; and the reverse DCF requiring 16.1% annual FCF growth for a decade ❌. Two green, one red, on a business that has actually grown FCF-adjacent earnings about 20% a year for ten years — which is precisely why the dissent is arguable rather than decisive.
Watch for
- An Earnings Growth Model whose growth input is the analysts' number. The haircut is what makes it worth reading.
- A reverse DCF hurdle close to the company's realised history. 16.1% required against 19-20% delivered is a much softer objection than it looks.
6. Deduct stock-based compensation everywhere, not only where it is convenient
The repeatable method
- Take SBC as a percentage of net income and test it against a fixed bar (<10%).
- Subtract the full charge from forecast free cash flow before running any DCF.
- Then do the same to the headline multiple, so the quick comparison and the model use the same earnings.
- Note the effect in points of multiple, so the size of the adjustment is visible rather than implied.
Here: the deduction is made inside the model — "The expected Free Cash Flow of the next 12 months equals
$63.4 million. We subtract the Stock-Based Compensation (
$6.6 million) to arrive at FCF in year 1 of
$56.8 million" — and
not made to the 32.5x headline multiple. The same asymmetry appears in the
May Arista dive and is corrected a week later in the
FICO dive, where restating 25.2x as 30.7x is the reason the deal is declined.
Watch for
- The adjustment being applied only to names you are inclined to reject. It is a discipline only if it also costs you names you like.
- Buybacks that are really offsetting dilution. With SBC at 14-16% of profit, a share count that is merely flat is a cost, not a return.
Methods distilled from the archived Compounding Quality post for personal study. Not investment advice.