ReportsMonitoringPortfolioMethodologyPricingBlogAboutStart for free
Back to blog

Method

What Is an Investment Thesis? A Worked Example on Costco

An investment thesis isn't a feeling or a target price. Three sentences about a business, each with a number and a date to check it, built on a real company.

8 min read
See a sample report

This is a method, not a recommendation. Nothing here, or anywhere else on Taufolio, is investment advice. Treat every example as a starting point for your own research.

An investment thesis is not a feeling and it is not a number for where the stock should trade. (If it were, half of Reddit would qualify.) It's three or four sentences about the business, never the stock price, each backed by either a document or a clearly named condition under which it stops being true, plus a date to check it. If your thesis is "it's going up, so I'm holding" or "cheap stock, good timing," you don't have a thesis. You have a mood in a blazer.

Let's build a real one together, on a company you've probably shopped at, so the difference is visible instead of just claimed.

What actually separates a thesis from a stock pick

A stock pick has one ingredient: direction. The price goes up or it goes down, and the whole exercise is guessing which. There's no way to check that except waiting, which makes a pick a bet, not an analysis.

An investment thesis is a different shape entirely. Instead of the price's direction, it describes the business mechanism underneath: where the profit actually comes from, what protects it, and what would have to happen for that mechanism to break. The price is an output of that mechanism, not a description of it, so two investors can hold the exact same thesis about a business and land on two different opinions about what it's worth today. That's why an Investment thesis on Taufolio never ends in a number next to the ticker. It ends in a list of business sentences you're supposed to keep watching.

The test is simple: can this sentence be disproven by a specific document on a specific date? "The company has a good product" fails that test. "Enterprise customers stay above 60% of revenue in the next quarterly filing" passes it, because a stranger with no opinion about the company can settle it from a table.

Three ingredients, or it isn't a thesis

A thesis worth the name stands on three legs, and each answers a different question.

  1. The money engine: one sentence on where the profit actually comes from, not the revenue.
  2. What protects it: why a competitor can't copy that mechanism in a year. This is the economic moat question, and a thesis with no answer to it is a description of a product, not a business.
  3. The breaking condition: a number and a date at which you'll admit you were wrong. Skip this and you have an opinion, and an opinion can never lose.

Three sentences is the floor at which a description of a business still holds together. Five is the ceiling at which you still bother rereading the list. Past five, the review starts looking like paperwork, and paperwork is the first thing anyone skips after the second quarter.

A worked example: the thesis on Costco's membership fee

Rather than leave this abstract, let's build a thesis on a business I've already picked apart on this blog: Costco's one-page company analysis.

Sentence one, the money engine. Costco earns on the membership card getting renewed, not the margin on the shelf. In fiscal 2025, membership fees brought in $5.3bn against $269.9bn of net sales, which is 2% of revenue but 51% of operating income. Delete that one line tomorrow and operating income falls from $10.4bn to $5.1bn while not a single pallet on the floor looks different. This sentence has a source: a specific page in the annual report.

Sentence two, what protects it. Buying scale and the membership card together build a loop where more members fund lower prices, and lower prices keep members renewing. The company never states it in those words, so this sentence is inferred, not sourced outright. What it does have is indirect evidence: gross margin on merchandise has sat near 11 cents on the dollar for years, even though Costco could easily raise it if it wanted to earn at the shelf instead of at the door.

Sentence three, the breaking condition. The proof that the loop still works is what renewals do after a fee increase, not sales growth on its own. The US and Canada fee rose about 8% from September 2024, and a year later renewals in that same region slipped from 92.9% to 92.3%. That is 0.6 of a percentage point, at most about $30m a year even if the whole $5.3bn fee line moved with it, well under 1% of operating income. Costco's own explanation is not the fee: more members signed up online, including through digital promotions, and they renew at a slightly lower rate. Small number, two readings, and the breaking condition is what tells them apart: "if US and Canada renewals fall below 92% in the next annual filing, sentence two's loop has its first real crack." A bounce back above 92.5% says it was the online cohort, not fee fatigue setting in.

Three sentences, three different origins, one shared property: each can be checked against a document, not a vibe. That's the entire difference between a thesis and a pick.

Bull case and bear case, read for the crack

With those three sentences in hand, the bull case and bear case write themselves, and they aren't two different theses, they're the same list read by someone hunting for where it breaks. A bear case on Costco doesn't say "what if the consumer pulls back," because that's a mood, not a testable sentence. It says: "if US and Canada renewals fall below 92% next fiscal year, sentence two loses its footing." Does that mean every thesis needs its bear case written on day one? Pretty much, yes. Labeling sentences like this - sourced, inferred, assumed - and turning them into a full bear case is the whole subject of how to break your own thesis before the market does. This piece is about writing the thesis; that one is about taking it apart before a quarterly filing does it for you.

The consensus view deserves a fair hearing here: Costco has grown for decades, so why bother with a breaking condition at all. Except growth rate and the direction of the membership loop are two different measurements. A company can grow revenue while quietly losing the thing that makes it resistant to a downturn, and the renewal rate reacts to that faster than revenue does.

Three ways a thesis quietly turns into an opinion

The most common failure is a thesis with no date. "Well positioned in the market" sounds like a sentence, but with no verification deadline it's a promise to think about it eventually, and that deadline can always slide another quarter.

The second is confusing a thesis with a target price for the stock. A target price is the output of a valuation model built on assumptions about a future nobody has seen. A thesis is a sentence about a mechanism that is either working or not working today. That's why two people can hold an identical thesis on Costco and land on very different opinions of what the stock is worth, and that's fine, because it's a different question.

The third is writing one sentence where you needed three. "Costco has a strong brand" sounds like a thesis, but it smuggles in several claims at once: customers know it, prefer it, pay to renew it, won't switch. Each of those has a different source and a different breaking condition, so one sentence has to become several before there's anything left to defend.

My own view is that the weakest link in most theses I've read, including my own, isn't a shortage of numbers. It's the missing third leg. Who enjoys writing the condition under which they'll admit they were wrong before anything has actually gone wrong yet?

Where the thesis lives after you write it

A written thesis is only half the job. The other half is coming back to it after every filing, which is exactly the part that's easy to postpone. The rhythm for doing that without reading every headline along the way is following a company on the filing calendar instead of the news feed.

In Taufolio, that second half doesn't rely on your memory. An Investment thesis comes out of a Full report already broken into assumptions, each with a measurable condition, and it gets revisited automatically after results and after events that could settle those conditions. You can see what a thesis split into conditions looks like in Monitoring.

Start with one company you think you know. Write the three sentences your view of it rests on, give each one a threshold, and pick a date for at least one of them. If a year from now you can't point to the document that settled it, you never had a thesis. You had a habit with a good explanation attached.

Frequently asked questions

A written reason for holding a stock, broken into a few sentences about the business, each backed by a document or a named condition and a date. It is not a feeling and not a price target - it is a list of things that have to stay true.
One load-bearing sentence about how the company makes money, one about what protects that mechanism from competitors, and one number-and-date condition that would prove the thesis wrong. This piece builds all three, live, on Costco's membership-fee model.
A stock pick ends at a direction: the price goes up or down. A thesis ends at sentences about the business that would have to explain that direction, so it can be checked against a document before the price ever moves.
No. A target price is a number about the share price; a thesis is a sentence about the business mechanism behind it. Two investors can share an identical thesis and land on two different numbers for the stock, because that number is an output of a model, not the thesis itself - which is exactly why Taufolio never publishes its own target prices.
Three to five. Fewer than that and you have not described a business; more than that and the review turns into a ritual nobody performs past the second quarter.
After every filing that touches one of your load-bearing sentences, which for most companies means once a quarter. A thesis with no date attached is not a thesis. It is a belief with a nice explanation.
  • research
  • methodology
  • portfolio
Share:XLinkedIn

Posts are produced with AI tools and go through editorial review by the Taufolio team before publishing. How we write and correct posts

Cookies and analytics

Essential cookies keep you signed in. Analytics runs only if you agree.