Most funds have a thesis. Fewer have one that two analysts, reading the same deck on the same day, would apply the same way. The usual version lives in a fundraising deck: a paragraph about a large shift, a few sectors, a stage and a geography. It persuades limited partners. It rarely tells an associate whether to take a first call.
Learning how to write an investment thesis that works in screening means writing two things: a short narrative that explains your view, and a set of criteria that turns that view into decisions. The narrative is for people. The criteria are for consistency. What follows covers both, with a worked example for a fictional fund, Northwind Seed.
What an investment thesis is for
A thesis does three jobs. It tells founders whether to approach you, which saves both sides time. It tells your team which deals to spend hours on and which to decline quickly. And it gives limited partners a basis for judging whether you did what you said you would do.
Fit is not the only thing investors weigh, but it is a real one. In the survey of 885 venture capitalists by Gompers, Gornall, Kaplan and Strebulaev, roughly half of firms named fit with the fund as an important factor in selecting deals, and 14% named it the most important. The same study found the average firm screened about 200 companies a year and closed about four. A clear thesis is what lets a small team decline the other 196 for consistent reasons.
A thesis also connects to what you told your investors. In the United States, the SEC rule that defines a venture capital fund for the adviser exemption requires the fund to represent to investors that it pursues a venture capital strategy, and limits assets other than qualifying investments to 20% of capital commitments. Your counsel decides how that applies to you, but it is one reason to write exceptions down rather than improvise them.
Start with the narrative in five sentences
Before criteria, write the view. Keep it short enough that every partner can say it from memory. Five sentences is a useful limit:
- 1What changes in the world create the companies you want to back?
- 2Which kinds of companies does that change favour, at which stage?
- 3Why will founders of those companies choose your fund over others?
- 4What does your fund bring after the cheque: expertise, customers, hiring, follow-on capital?
- 5What are you deliberately not doing, even when the deal is good?
The last sentence is the one most funds skip, and it is the one that makes screening faster. Saying "we do not back hardware, consumer apps or companies raising above a certain round size" removes more ambiguity than any amount of language about large markets.
For Northwind Seed, the narrative might read: software that automates regulated back-office work is becoming possible as models improve and compliance rules are published as data. We back the first institutional round of business software companies in that space. Founders choose us because our partners ran operations at regulated companies. We introduce first customers from that network. We do not back consumer products or companies past their first institutional round.
How to write an investment thesis as criteria
Now turn the narrative into criteria. Separate them into four kinds, because each behaves differently when a deal does not meet it.
| Part | What it means | Northwind Seed example | What happens when a deal misses it |
|---|---|---|---|
| Hard rules | Conditions every deal must meet | First institutional round; business software; company can take investment from our fund's jurisdiction | Mismatch is surfaced for a person to review and record a reason |
| Weighted preferences | Signals that make a deal more attractive, each with a weight | Regulated buyer (high); founder has worked in the industry (high); early revenue (medium) | Lowers attractiveness but never decides alone |
| Exceptions | Named cases where a hard rule may be waived, and who can approve | A second-round company if the lead partner sponsors it and the budget for exceptions allows | Needs a named approver and a written reason |
| Open questions | What you cannot tell from a deck and need to ask | Who signs the purchase order? Is there a compliance deadline driving demand? | Becomes a question for the founder, not a failure |
Write each criterion so someone else could check it. "Strong team" cannot be checked. "At least one founder has worked inside the type of organisation the product sells to" can. "Large market" cannot be checked. "A bottom-up estimate of at least a stated number of potential customers, with the assumptions shown" can.
Keep the list short. Five hard rules and six to eight preferences is plenty for most seed funds. A long list encourages box-ticking and hides the few things that really decide a deal.
Handle unknowns, mismatches and overrides
A thesis is only as useful as the way your team applies it to incomplete information, and early-stage information is always incomplete. Agree three rules before anyone screens a deck.
- Unknown is not a fail. If a deck does not say where the company is incorporated, the criterion is unknown. It becomes a question in the founder's follow-up, not a reason to decline.
- Mismatch goes to a person. A deal that clearly falls outside a hard rule is surfaced for review. Most will be declined quickly, but a person decides and records why.
- Overrides need a reason. When a partner wants to proceed despite a mismatch, they write one or two sentences explaining why. Over a year, those reasons show you where the thesis is out of date.
These rules matter more as more of your screening is assisted by software. A tool that treats missing data as failure will quietly reject companies whose founders wrote short decks. Our guide on how to screen startup applications covers applying criteria at volume.
Version it, share it and review it
- Date every version. When you change a rule, keep the old one so you can see which deals were screened under which thesis.
- Publish the founder-facing part. A short page on what you back, what you do not, and what you ask for saves founders an application they did not need to write.
- Review on a schedule, such as every six months, using the override reasons, the deals you passed on that later raised elsewhere, and the deals you backed.
- Use different rubrics for different strategies. If you run a seed programme and an opportunity fund, they need separate criteria, not one compromise.
- Avoid common mistakes: a thesis that only describes a trend, criteria nobody can check, no stated exclusions, and changing the rules mid-cycle without telling the team.
Our investment thesis template gives a structure to fill in, and the thesis builder turns your answers into rules, preferences, exceptions and questions you can export.
Where Delsivo fits
Delsivo uses the same structure. In the pilot release, a fund writes its thesis as versioned hard rules, weighted preferences, exceptions and questions, with different rubrics by strategy or stage. When a deck arrives, each criterion is shown with the rule, the value found, the slide it came from and the reasoning. Results are Met, Mismatch or Unknown, and Unknown becomes a question for the founder.
Thesis fit is shown separately from evidence quality and diligence readiness, so a well-fitting deal with thin evidence does not look the same as a well-evidenced one. Overrides need a reason, and re-running a rubric creates a new run while keeping the old one. Delsivo suggests and explains. People decide every pass and every next step. The thesis screening page has the details.
Sources
Delsivo publishes this guide and makes software for venture funds, so we have a stake in the topic. Facts about other companies, research and regulations come from the sources listed, on the dates shown, and can change. Corrections are welcome through our contact page.