Track outcomes the fund cares about, plus the few leading indicators that predict them. Count requests and intros, then measure what they produced: meetings, hires, customers and capital raised. Add response time and founder reach. Define each metric in writing, capture the data automatically, and drop any number nobody acts on.
When I ask platform leads what they track, the most common answer starts with a pause. Then: "Well, founders tell us we're helpful." I get it. It's true, and it's the easiest thing to collect. But it won't carry you through a budget meeting, and it can't answer the partner who asks what changed because platform exists.
The opposite problem is real too. A fund ops lead told me they quietly stopped tracking one portfolio metric because nobody ever looked at it. That's the right instinct, and it applies to platform metrics as well. A metric is worth tracking only if someone makes a decision with it.
Use three layers: activity, outcome and fund impact
Most platform dashboards stop at activity. That's the layer that's easiest to count and least convincing. Add two more on top.
| Layer | What it answers | Examples | Who cares most |
|---|---|---|---|
| Activity | Are we doing the work? | Intros made, requests handled, events run | The platform team |
| Outcome | Did the work produce something? | Intros that became meetings, pilots, hires or investments | Partners |
| Fund impact | Did it help the fund? | Deals won with platform evidence, rounds supported, LP interest | GPs and LPs |
Activity still matters. It tells you where founders need help and whether the team has capacity. But on its own it's a measure of effort, and partners stop caring about effort the moment software does half of it.
The fund-impact layer is the hardest to measure, and you won't get clean numbers there. That's fine. A few well-sourced stories each quarter ("platform's case studies came up in two competitive deals we won") count as evidence.
A starter metric set
Here's an illustrative set for a platform team covering about 30 companies. Pick five or six, not all of them.
| Metric | Definition | Where the data comes from |
|---|---|---|
| Intro volume by type | Customer, investor and candidate intros made in the period | Intro email threads |
| Intro conversion | Share of intros that led to a second meeting within 30 days | Follow-up replies, calendar |
| Intro outcomes | Intros that became a pilot, contract, hire or investment | Founder replies, public announcements |
| Time to first reply | Median hours from a founder request to a useful first response | Email and Slack timestamps |
| Time to shortlist | Days from a senior role going live to candidates in the founder's inbox | Job posts, sent emails |
| Founder reach | Share of companies with a real conversation with platform in the last 60 days | Calendar, email |
| Asks routed | Asks found in updates and routed to someone who can help | Founder updates, intro threads |
| Value-add log entries | Outcomes logged with a source link and, where possible, a founder quote | The log itself |
Founder reach is the one I'd add first if you track nothing today. It catches the quiet companies. A founder going silent rarely means nothing is happening, and the companies that need help most are often the ones not asking for it.
Write a definition for every metric
A metric without a definition turns into an argument the first time the number looks bad. Write one short definition per metric, and share it with partners before you report anything.
A template:
- Name: Intro conversion
- Question it answers: Are our intros relevant enough to get a second meeting?
- Counts: Customer and investor intros where both sides opted in first.
- Doesn't count: Mass intros, event introductions, intros where either side declined.
- Source: The intro email thread, plus a follow-up email two weeks later.
- Owner: The platform lead.
- Reviewed: Monthly.
- Decision it informs: Which intro types to do more of, and which contacts to stop using.
That last line is the test. If you can't name the decision, drop the metric.
Capture the data without anyone logging anything
Here's where most platform metrics die. The plan depends on someone filling in a spreadsheet after every intro, and by week six the spreadsheet is a quarter behind. People who log by hand forget. It isn't laziness. The work is happening in email and Slack, and the spreadsheet is somewhere else.
So capture from where the work already happens:
- Intros: the double opt-in email thread is the record. Pull who, what and when from it.
- Outcomes: a short follow-up two weeks after each intro asking "did this go anywhere?" The reply is the data.
- Response time: timestamps already exist in your inbox and Slack.
- Founder reach: calendars already know who you met.
And keep a link from every number back to its source. When a partner asks "which intros?", you click through to the thread instead of rebuilding the answer from memory.
The same rule holds for founder-side data. Don't send founders a form asking how helpful you were. Ask one question in the existing thread after a big piece of help, and let them reply however they like. Cura does a version of this for portfolio data: it pulls signal from email, board decks, LinkedIn, X and news into one feed, and creates follow-up tasks tied to each company, so the record builds as the work happens.
Retire metrics on a schedule
Once a quarter, go through your metric list and ask of each one: did anyone make a decision with this? If not, cut it or change it. Platform teams pick up metrics over time and almost never drop them, so the dashboard grows until nobody reads it.
Goals drift too. What the fund needed in its investing period isn't what it needs during a raise. When the fund's priorities shift, your metrics should shift with them. If you haven't agreed goals with the GP yet, start there. I wrote about that conversation in going from support function to value engine.
Why isn’t “founders find us helpful” a good platform metric?
It measures goodwill, not results. Founders are polite, surveys reach the friendliest ones, and “helpful” can’t tell a partner which programs to fund or cut. Keep the sentiment, but pair it with outcomes: the intro that became a customer, the hire that closed, the round that came together faster.
There's a second problem. "Helpful" scores barely move. If every founder rates you a four or five, the number can't tell you whether this quarter was better than last.
Better questions get closer to outcomes. After a significant piece of help, try asking in the same thread:
- "What would you have done if we hadn't helped with this?"
- "Did anything come of the intro?"
- "Would you be up for telling another founder about this?"
The first one shows whether the help changed anything. The third one builds your reference bench as you go.
How often should platform metrics go to partners?
Send a short monthly note and hold a deeper quarterly review. Monthly keeps platform visible and surfaces problems early. Quarterly is when you compare results against the goals you agreed with the GP and decide what to change. Save the annual view for planning and LP materials.
A cadence that works for most teams:
| Cadence | Audience | What's in it |
|---|---|---|
| Weekly | Platform team | Open requests, signals to act on, intros awaiting follow-up |
| Monthly | Partners | Progress on goals, one or two outcomes, one miss, one ask |
| Quarterly | Partners | Full metric review, what to start and stop, goals check |
| Annually | GPs, LP materials | Year in review, best case studies, plan for next year |
| On demand | A deal partner | Relevant outcomes and references for a live competitive deal |
The on-demand row matters more than it looks. The best test of your metrics is whether you can answer "how have we helped companies like this one?" in minutes when a partner is in a competitive round. If that answer takes three days, the data isn't working for you yet. For more on the expectations behind these numbers, see how AI is changing the platform team, and for why the measuring matters at all, the value-add gap.