Your investor update is a data feed now

Portfolio monitoring used to mean an analyst reading your update and typing three numbers into a spreadsheet. Sometimes months late.

That job is now done by software that reads board decks, financial statements and founder emails, and extracts the numbers itself.

One fund, Davidovs Venture Collective, launched a $75 million fund and did without five analyst roles. Agents cover sourcing, diligence and portfolio monitoring.

Your update stopped being a letter

It is an input to a database. The prose gets read once by a person, if you are lucky, and parsed every time by something else.

The numbers inside it become rows. Those rows sit next to the rows from your last six updates.

Which means the interesting question is no longer whether this month reads well. It is whether this month agrees with the other five.

Drift is the thing that gets caught

Nobody sets out to mislead their investors. Drift happens by accident, in three ordinary ways.

You change a definition. Revenue meant bookings in March and collected cash in June, and you did not flag the change.

You change the window. Growth was month on month while it was strong, then quietly became quarterly.

Or a metric simply disappears. You stopped reporting churn four updates ago, around the time it got worse.

A person reading one update misses all three. A system holding all six does not.

Write it so it parses

This costs nothing and takes one hour to set up once.

Put the numbers in a fixed block, same order every month, same names every month. Then write whatever you want underneath.

The prose is still where you build the relationship. Ask for intros, admit what went wrong, say what you learned. That part is for a human.

The block above it is for the machine, and being generous to it is not a compromise.

The prompt

Run this over your last six updates before you send the seventh. It does what their system does.

You are auditing a series of investor updates for consistency, the way a
portfolio monitoring system would.

Here are my last six updates, oldest first:
[PASTE ALL SIX]

Produce, in plain language:

1. A table with one row per update and one column per metric mentioned.
   Leave a cell blank where the metric was not reported that month.
2. DEFINITION CHANGES: any metric whose meaning appears to have shifted,
   with the two quotes side by side.
3. DROPPED METRICS: anything reported at least twice and then absent,
   with the month it stopped.
4. WINDOW CHANGES: anywhere a comparison period changed, for example
   month on month becoming quarter on quarter.
5. CONTRADICTIONS: any two statements across the six that cannot both
   be true. Quote both.
6. The three questions an investor would ask after reading all six
   together, ranked by how uncomfortable they are.

Rules: quote directly, never paraphrase into something kinder. If a
number is ambiguous, list it under AMBIGUOUS rather than picking a
reading. Do not comment on tone or writing quality.

What to do with what it finds

Fix the format going forward, and deal with the history in one line.

If a definition changed, say so in the next update. Give the old number and the new number for the same month, once.

That single line costs you nothing today. It removes a question that would otherwise surface during diligence, eighteen months from now, with a term sheet on the table.

The founders who get hurt by this are not the ones with bad numbers. They are the ones whose numbers cannot be reconciled.

The update consistency audit is on the resources page. Free, no email required.