A VC firm looks at 101 deals on average for every investment it makes, according to the survey by Gompers, Gornall, Kaplan and Strebulaev of 885 investors (run in 2015 and 2016). The investment gets a committee memo. The other hundred leave one word in the CRM, if they leave anything, and that word is useless the day the company comes back.
What happens when the company comes back
Take an invented but ordinary case. A company you passed on in March 2025 sends a new deck in October 2026. The analyst who read the first one has moved to another fund. The CRM says “too early”. Nobody can tell whether that meant too little revenue, no CTO, or a market the partner didn’t want to look at that year.
The team can redo the analysis from scratch, or pass again on a memory, for a reason that may no longer hold. It can’t tell what has changed, because it doesn’t know what it was waiting for.
The same problem shows up across the firm. In that survey, only one investor in ten says they quantitatively analyze their past decisions. Gompers et al., How Do Venture Capitalists Make Decisions?, 2016. A firm with no written reasons has nothing to analyze on its passes, and passes make up most of its decisions.
Bessemer Venture Partners does the opposite, in public. Its Anti-Portfolio page tells the stories of the companies it let go. Byron Deeter met Atlassian in 2006 and noted that the founders didn’t want to go public, and in 2010 the firm found the $400M valuation a bit rich. The page is meant to be funny. It also shows that a pass written down with a fact and a date can still be read twenty years later.
“Too early” doesn’t say when
A one-word reason files the deal without explaining it. The useful version fits on two lines and makes sense to someone who never saw the deal, including a person who joins the firm two years later.
Three invented examples
“Too early” becomes: no revenue, three free pilots, none converted. Revisit if they sign five paying customers.
“Team” becomes: nobody technical, product built by an agency. Revisit if a CTO joins.
“Market too small” becomes: we estimate spend at €40M a year in France. Revisit if the company sells elsewhere.
Four fields are enough
A pass note holds the reason, picked from a short list, the fact behind it with its source and date, what would change your mind, and the name of the person who decided.
The third field takes the most effort, because it forces you to say what you would have wanted to see. Write something a third party could check: “€1M ARR” can be checked, “more traction” can’t. The condition can also be an event, such as a CTO hired or a license granted, or a plain date. If nothing comes to mind after thirty seconds, the deal is probably out of thesis and there is nothing to wait for.
Template to copy
Reason: [one only, from the fund’s list]
Fact: [figure or finding, deck page or link, date]
Revisit if: [figure, event or date]
Decided by: [name, date]
Invented example. Reason: too early. Fact: no revenue, three free pilots (deck, page 9, March 2026). Revisit if: five paying customers. Decided by: Claire, March 12, 2026.
A short list, specific to the fund
We suggest eight reasons at most. Past that, people end up picking “Other” and the count stops meaning anything. A starting point: out of thesis, too early, team, market, traction, product, valuation, competes with a portfolio company. A deep tech fund will add technical risk, an impact fund its exclusion criteria.
Keep deals closed without a judgment apart: the duplicate, the founder who withdraws, the round that closed without you, the founder who stopped replying. We made this mistake in our own data, where a deal archived as a duplicate counted as a pass. The day you look at why the fund says no, those lines skew the count.
Separate “out of thesis” from the other reasons too. A pass on the team says something about the company, and a pass on thesis says something about the fund. When the thesis changes, those are the deals to reopen first.
One minute, when you pass
The reason gets written when the rejection goes out, by the person who read the deal. A week later they remember their conclusion and have forgotten the number that tipped it.
Write facts. In Europe, a note about a named founder is personal data, and the founder can ask to see it under the GDPR right of access. “Founder not convincing” is hard to defend. “Said €2M ARR on the call, €1.1M in the deck” is easy to defend, and more use to whoever reads it next.
The condition also helps in the message to the founder. “Come back when you have five paying customers” gives them a target, and they’ll return with the number instead of sending the same deck six months later.
Reread twice a year
Half an hour every six months is enough. Start with the passes whose date has come up or whose condition looks met: a search for a funding announcement or a look at the company register often answers it in two minutes. Then count the reasons. If half of the passes in that period are out of thesis, the fund is getting the wrong deals, and the answer lies in sourcing or in how it presents its thesis. If one reason dominates for one person, talk to them about it.
Keep one limit in mind. Maxime Bonelli (HEC Paris) studied VCs who adopt data tools. They get better at avoiding failures and become less likely to pick the rare big winners, because they move toward what has already been tried. Bonelli, Data-driven Investors, 2023. A reason describes what a team thought of a company on a given date. Never use it to reject a company automatically because it looks like another one.
In Casian, archiving a deal means choosing a reason, with an optional explanation and, for “Too early”, a follow-up date. The reminder appears on the Desk without putting the deal back in the pipeline. To see what the screening picks up on one of your decks, try the free screening, no account needed.