Actionable insights — Is L'Oréal an interesting stock?

1. Separate the two ways a wonderful business can be uninvestable: price and growth

1. Finish the quality assessment and score it before opening any valuation work, so the two verdicts cannot contaminate each other.
2. Then ask two distinct questions. Is the business priced above what it is worth? And separately: is the business still capable of growing fast enough to earn a return at any sensible price?
3. A price problem is temporary and gives you a watchlist entry. A growth problem is structural and may never resolve — a business at the law-of-large-numbers ceiling does not grow out of it.
4. Say which of the two is binding, in the verdict itself, so the follow-up action is obvious.

• A "too expensive" verdict that is really a growth verdict. If the required multiple to make the numbers work is far below the company's own history, the model is telling you the growth is gone, not that the market is wrong.
• Analyst forecasts sitting exactly on your threshold. A 7.0% long-term EPS estimate against a 7% bar is a fail dressed as a pass.

2. Test whether a consumer moat is owned or rented, using the advertising line

1. Take advertising and promotion spend as a percentage of revenue and read it as the annual maintenance cost of the brand.
2. Ask what happens to volumes if that spend is halved for two years. A genuinely owned brand survives it; a rented one does not.
3. Cross-check against gross margin trend: if margin holds while ad spend as a share of revenue rises, the brand is getting more expensive to defend.
4. Compare the number to peers in the same category, since the absolute level means little on its own.

• The digital shelf. The stated erosion mechanism is specific: smaller competitors "don't have to battle for expensive shelf space anymore", which removes the distribution half of the scale advantage while leaving the advertising bill intact.
• A moat score that ignores the question the same paragraph just asked.

3. Run three valuations and let them disagree — the split tells you which risk you are taking

1. Method one: forward PE against the company's own ten-year average. This tests sentiment only.
2. Method two: an earnings-growth model — expected EPS growth + dividend yield ± the annualised effect of the multiple reverting to a stated target. This gives an expected annual return you can compare with a hurdle.
3. Method three: a reverse DCF — solve for the FCF growth the current price implies, then compare it against the growth actually delivered over the last decade.
4. Record the verdict of each separately. Agreement is a signal; disagreement tells you exactly which assumption is doing the work.

• Relying on the cheapest of the three. The multiple test is explicitly called "a shortsighted method to give a quick indication."
• A reverse DCF that is built properly: expected FCF less stock-based compensation plus growth capex — €7,350.0m − €248.0m + €54.0m = €7,156.0m in year one. Skipping the SBC deduction flatters every result.

4. Publish the failed thresholds instead of arguing them away

1. Fix the thresholds before you look at the company: gross margin >40%, ROIC >15%, ROE >20%, interest coverage >15x, goodwill/assets <20%, CAPEX/sales <5%, net margin >10%, FCF/net income >80%, SBC <10% of net income.
2. Mark each pass or fail mechanically, and write the fails down next to the passes.
3. Score each of the 15 sections out of 10 and add them into one Total Quality Score, so a strong section cannot silently rescue a weak one.
4. Then decide — with the failures visible, not summarised away.

• Goodwill as a tell. A 23.4% goodwill/assets ratio on a "brand" company means a meaningful share of the portfolio was bought at a premium rather than built — and the ROIC denominator carries that premium, which is part of why ROIC is only 13.9%.
• Scores that never fall below 6.5. A range that narrow means the scoring is descriptive rather than discriminating.

5. Convert every "no" into a multiple, then into a share price

1. State the multiple at which the business would clear your return hurdle, given your growth estimate rather than the market's.
2. Convert it into a share price and publish both numbers side by side with the current one.
3. Set the alert at that price; do not revisit the name on a calendar.
4. Keep the quality work — it does not need redoing when only the price changes.

• Whether the target multiple is defended. 20x is asserted rather than derived here — the earnings-growth model would need roughly that multiple to clear a 10% return, but the link is not spelled out.
• A target on forward earnings that themselves get revised down. €271 at 20x is only a real level while the forecast holds.

6. Read management tenure and family ownership as evidence, not as colour

1. Count the number of chief executives across the company's whole life and divide — then compare against the market norm.
2. Check whether the current CEO is an insider promotion with a long internal history, and how much stock they personally own.
3. Identify the largest shareholder and whether the stake is inherited, concentrated and board-represented — inherited concentrated stakes cannot be sold quickly, which is what makes them align with the long term.
4. Only then classify the company: an Owner-Operator label should follow from the ownership facts, not from the narrative.

• Long tenure at a company whose end market is changing. Stability is an asset against disruption and a liability against reinvention — and the risk named here (digital-native niche brands) is exactly the kind incumbency handles badly.
• Insider ownership counted as a percentage without asking whether the holder can or would ever sell.

7. Screen on what will not change, and price the mature end market honestly

1. Ask Bezos' question in the form used here: what will not change in ten years? Build the durability case on that, not on the growth story.
2. Then separate durability from growth explicitly — a low risk of disruption is a reason to hold a business a long time, not a reason to pay a growth multiple for it.
3. Quantify the market's own growth rate independently of the company's (here 6.7% a year to 2030) and check the company's forecast against it.
4. Where the two are similar, accept that the investment case must come from price rather than from share gains.

• Durability arguments used to justify a multiple. This is the same tension the August 2026 Lindy series runs into, where survival is argued as a valuation input and no price is ever attached.
• Big-number tailwinds (600 million new customers) that do not appear in the two-year revenue forecast.