Actionable insights — CMG: The Next Constellation Software?

1. Score every candidate on the same 15 questions — and publish the number

1. Work through a fixed sequence: (1) do I understand the business model, (2) is management capable, (3) is there a sustainable competitive advantage, (4) is the end market attractive, (5) what are the main risks, (6) is the balance sheet healthy, (7) how capital-hungry is it, (8) is it a great capital allocator, (9) how profitable is it, (10) how much stock-based compensation, (11) did it grow in the past, (12) does the future look bright, (13) is the valuation fair, (14) how did owner's earnings evolve, (15) did it create shareholder value.
2. Score each of the fifteen and roll them into one Total Quality Score out of 10, so different companies can be compared on the same scale months apart.
3. Answer the questions in that order — the qualitative ones first, valuation second-to-last. Valuation never rescues a business that failed steps 1–12.

• Which specific steps failed — the composite hides them; a name that scores 8+ purely on moat and balance sheet while failing every growth step is a different bet than one that passes evenly.

2. Use hard numeric thresholds so "quality" isn't a judgement call

1. Attach a published threshold to every quantitative step and mark each pass or fail explicitly.
2. Apply the same numbers to every candidate, and record the failures rather than arguing them away.

• A candidate that passes the balance-sheet and margin thresholds but fails every growth threshold — then the entire case rests on the turnaround being real.

3. Judge a new CEO by the delta, not the CV — "words are cheap, actions speak louder"

1. Read every shareholder letter the CEO has written and note the words that recur — they reveal what is actually being optimised for.
2. Then measure the operating record from the date they arrived, against the record before, rather than looking at 5- and 10-year averages that mix two different companies.
3. State the limitation honestly: a short tenure means there is no track record, only a promising delta.

• The since-arrival numbers rolling over; and whether the improvement came from price, mix, or acquisitions rather than from the operating discipline the letters claim.

4. Screen for the cloned playbook — proven model, proven people, smaller company

1. Start from a business model already proven to compound (here: buying vertical market software and leaving it decentralised).
2. Look for a much smaller company where the people from the proven model have shown up — as chairman, largest shareholder, or head of M&A — since the playbook travels with the operators, not the org chart.
3. Sanity-check the runway by comparing scale: the imitator's small deal count against the original's, so the reinvestment opportunity is decades long rather than nearly exhausted.
4. Look for the same alignment devices being installed, not just the same strategy talked about.

• Whether the acquisitions keep coming at disciplined prices — one or two deals is a claim, not yet a machine; and key-man risk, which Slegers names as the single biggest risk here.

5. Read the compensation redesign as the hardest evidence of intent

1. Check what executives are measured on. A shift from revenue growth alone to ROIC changes which deals get done.
2. Check the form of the bonus: cash the employee must use to buy shares on the open market puts real money at risk, unlike granted RSUs that arrive free and dilute holders.
3. Treat a dividend cut in favour of reinvestment at high returns as a positive, not a negative, when ROIC exceeds the cost of capital.

• Proxy-statement changes as an early signal: they usually precede the change in behaviour by a year or more.

6. Take an explicit position on the consensus estimate — and say which way

1. Write down what the sell side expects, in numbers, before forming your own view.
2. If you disagree, say so and list the specific reasons the estimate is wrong (returns on capital, capital-deployment capability, management change) — the disagreement is the thesis.
3. Recognise the asymmetry: buying a business whose consensus long-term growth is negative means the price already embeds the pessimism.

• Estimates that extrapolate the pre-turnaround period; and the first quarters that either confirm or kill the disagreement.

7. Triangulate value three ways — and require all three to agree

1. Multiple vs its own history: compare today's forward PE with the 10-year average multiple for the same company.
2. Earnings-growth model: expected return = EPS growth + dividend yield ± multiple change. Use conservative inputs — and adjust the yield for a payout you expect to change.
3. Reverse DCF: instead of forecasting, solve for the free-cash-flow growth rate the current price requires to deliver your hurdle (10%/yr), then ask whether that number is easy or heroic.
4. Only act when all three point the same direction; a single cheap-looking method is not enough.

• A reverse-DCF requirement that sits above the company's own historical growth — that's the price doing the work, not the business.

8. Let the drawdown be the entry, once the score is already high

1. Do the quality work first and independently of the price action, so the score is not a rationalisation of a falling chart.
2. Then look at the drawdown and long-run compounding record together: a −35% year against a 16.3%/yr CAGR since IPO is a price event, not a business event.
3. Buy the discount only where the reverse DCF says the price now demands less than the business has historically delivered.

• Whether anything in steps 1–12 actually changed during the drawdown; if the renewal rate, the moat and the acquisition pipeline are intact, the fall is multiple compression.