The Far-Analogy Framework Transplant
Take an analytical framework from a distant field and run it across your own companies.
- Difficulty
- Moderate
- Time to result
- ~weeks to results
- Steps
- 6
- Confidence
- 90%
All the analysts at Gurley's firm shared a floor, so the food analyst, the telco analyst and the tech analyst sat together, and the firm ran efforts to teach the analysts a common framework. Gurley became close with Michael Mauboussin, the food analyst, who had just read a set of books on return on invested capital — the Stern Stewart material and McKinsey's Valuation — and was proselytising it through the analyst group. Gurley, in his words a sponge at that point, took the framework and ran it across all of his companies. Dell stood out like a sore thumb, with ROIC numbers versus the rest of the industry that were night and day, roughly twenty to one. The result was so extreme that Mauboussin initially did not believe it was true and they went back into the numbers. Two lucky things coincided: Dell had stumbled on a bad options trade and laptops catching fire, so the stock was in the ditch at six times earnings. Armed with a framework nobody else in tech was applying, Gurley could see a massive competitive advantage that the price did not reflect. He went to a strong buy on a broken stock and it went up 100x from there. Later in the episode he generalises the pattern via David Epstein's Range and Holyoak's work on far analogies.
Origin
Gurley picked up return on invested capital in the mid-1990s from Michael Mauboussin, then the food analyst on the same research floor, who was spreading the framework after reading Stern Stewart and McKinsey's Valuation. Gurley applied it to PC hardware, where nobody was using it.
Core principles
- 01The biggest breakthroughs in science come disproportionately from people who changed disciplines and brought a different mental framework with them.
- 02Borrowing an idea from far away is an actual intellectual skill, which UCLA's Holyoak studies under the name far analogies.
- 03A framework that is standard in one field is often unknown in an adjacent one, which is where the edge lives.
- 04Transplanting is cheap: run the framework across your entire coverage universe and look for the outlier, not the average.
- 05When the outlier is so extreme that people refuse to believe the numbers, that is a signal, not an error.
- 06Proximity to other disciplines is engineerable — shared floors, multidisciplinary institutes, unfamiliar podcasts.
How to run it
- 1
Engineer proximity to other disciplines
Deliberately spend time where practitioners from unrelated fields mix. Gurley's floor put him next to a food analyst; his Santa Fe Institute involvement puts biologists next to physicists next to epidemiologists.
Pro tip Consume a podcast or book from a field with no obvious relevance to yours on a regular cadence.
- 2
Ask what framework they are currently excited about
When you meet someone from another discipline, ask what analytical tool is changing their work. Be a sponge rather than an evaluator on the first pass.
Pro tip Natural learners proselytise freely — Gurley attributes Mauboussin's sharing to genuine curiosity, not reciprocity.
Watch out Category-rejecting a framework because it comes from an unglamorous industry is how you miss it.
- 3
Learn the framework at source
Read the primary material rather than the summary. Mauboussin came in via the Stern Stewart books and McKinsey's Valuation, not a secondhand explanation.
Watch out A half-understood transplant produces confident errors that are worse than no framework at all.
- 4
Run it across your entire universe at once
Do not test the framework on one favourite case. Apply it to every company or case you cover so that outliers surface by comparison.
Pro tip The comparison across the whole set is what makes a 20-to-1 gap visible.
- 5
Verify the outlier before you act on it
When one result is absurdly extreme, go back into the underlying numbers with someone who knows the framework better than you do. Mauboussin did not believe the Dell result until they re-derived it.
Pro tip Disbelief from an expert is a good reason to double-check and a good sign you have found something.
Watch out Extreme outliers are usually data errors; assume that first and disprove it.
- 6
Publish where the framework and the price disagree
Act hardest when your transplanted analysis says quality and the market says broken. That gap is the entire return.
Pro tip Being right is not enough — you have to be right and contrarian.
In the wild
Gurley took the return-on-invested-capital framework from the food analyst and ran it over his entire PC coverage. Dell's ROIC was roughly twenty to one versus the industry, while the stock traded at six times earnings after an options-trade stumble and laptops catching fire.
→ He went to a strong buy on a broken stock; it went up 100x in the public markets and the call put Bill Gurley on the map with the buy-side community.
At a Santa Fe Institute session on North American electric grid failures, the presenting professor concluded that the best solution was smaller communities that are loosely coupled. Gurley walked up afterwards and told her she had just explained why the euro is a bad idea, because it coupled its members too tightly.
→ A framework from grid engineering became a usable lens on monetary unions and on distributed computing systems, where too little coupling loses scale and too much creates global failure modes.
Common mistakes
Deferring to the specialists in the field
Gurley notes the constant refrain online is that you should leave a field to the people in it. But if you study science and history, the biggest breakthroughs come from people who moved over with a different mental framework and saw things differently.
Testing the transplant on a single favourite case
Running the framework on one company you already like produces confirmation, not discovery. The Dell result was only visible because the framework ran across the whole coverage universe.
Believing an extreme outlier immediately
The first response to a 20-to-1 result should be disbelief and a return to the raw numbers, exactly as Mauboussin's was. Skipping that step is how a data artefact becomes a strong buy.
Is it for you?
Best for
Researchers, investors and operators who need a differentiated view and have access to practitioners outside their own discipline.
Not ideal for
Situations where the transplanted framework's assumptions plainly do not hold in the new domain and would produce confident nonsense.
From the transcript
“I took the framework and ran it on all of my companies”
“a sore thumb with ridiculously high roic numbers versus the rest of the industry like night and day like 20 to one wasn't even close”
“we had discovered if you will through this framework that they had a massive competitive Advantage because of this return on invested Capital thing and…”
“there's a professor at UCLA named Hollybrook who did a piece on something he calls far analogies where he views it as an intellectual skill…”
From the episode
#651: Legendary Investor Bill Gurley on Investing Rules, Finding Outliers, Insights from Jeff Bezos and Howard Marks, Must-Read Books, Creating True Competitive Advantages, Open-Source Strategies, Adapting Mental Models to New Realities, and More