Investment thesis template: claims, evidence and review triggers
An investment thesis template should explain what must happen in a business, why it might happen, and what evidence would make you change your mind. A company description is useful context, but it does not provide those tests. Keep the first version short enough that a later reader can identify exactly which claim changed.
Start with a mechanism, then attach observations
Write one causal sentence: “If X persists, Y should improve because Z.” Separate observations already in a report from assumptions about the future. Record the reporting period, publication date and relevant population beside each observation. “Revenue grew” is incomplete if it refers to an acquired business while your thesis concerns existing customers.
Use the annual report to locate evidence. Investor.gov identifies business, risks, MD&A and audited financial statements as distinct 10-K sections. Management’s explanation belongs in the claim column until you corroborate the mechanism. The template below is an editorial research tool, not an official reporting requirement.
Worked example: a hypothetical maintenance business
Assume a maintenance company reports 100 units of revenue from existing contracts and 20 from installations. Your thesis is that renewals can stabilize cash generation as installations slow. This is different from predicting that the whole company will grow. The essential assumptions are customer retention, service costs and collection timing.
| Thesis card field | Hypothetical entry |
|---|---|
| Claim | Renewals can offset weaker installations over two renewal cycles |
| Supporting observation | Existing-contract revenue is 100; source page and period must be attached |
| Missing evidence | Renewal cohort definitions and service cost allocation |
| Competing explanation | Price increases disguise customer losses |
| Warning | Receivables rise while service revenue grows |
| Invalidation | Comparable renewal cohorts shrink across both observed cycles |
| Next review | After the second cycle is reported, allowing disclosure lag |
The two-cycle choice is an example of a research rule, not a universal threshold. If annual renewals are concentrated in one month, two calendar quarters will not test the same thing. Specify whether price increases count as success and whether acquisitions are excluded before seeing the result.
Check the card before using it
Every critical assumption needs either an observation route or an explicit “not currently observable” label. Follow the source link, verify units, and check whether the latest filing changes a definition. Then ask whether the competing explanation could produce the same headline numbers. If it could, add a discriminator such as customer counts or service gross margin; do not add another supportive headline.
Avoid expanding the thesis until every business strength becomes essential. Three necessary assumptions are easier to test than twelve loosely connected reasons to like a company. This card is unsuitable for a precise valuation conclusion without a separate valuation model. It also cannot turn incomplete disclosures into knowledge.
A further check on the example
Before the maintenance company’s next review, write down how you will treat contract migrations. If a customer moves from maintenance to a bundled installation package, a lower maintenance line might reflect classification rather than departure. Keep the original cohort, track the migration where disclosed, and avoid treating missing customers as confirmed cancellations. The final card should distinguish a failed retention assumption from a failed measurement proxy; those findings require different revisions and should remain visible to the next reviewer.
FAQ
Should a thesis contain a target price?
It can exist without one. A business mechanism and its tests are a separate research deliverable from valuation and a portfolio decision.
What if the company never discloses the required metric?
Record the gap and seek a defensible proxy with stated limitations. If no useful observation exists, reduce confidence or narrow the claim rather than inventing a measurement.