SaaS Metrics for Early Founders: Only Four Actually Matter

You wired up analytics in an afternoon. You have a dashboard with fourteen charts. Signups are up 20% this month, page views are climbing, and you have 300 GitHub stars.

You still have no idea what to do on Monday.

A metric matters if a change in it would change what you do next week. Everything else is decoration, however interesting. At your stage there are four numbers that pass that test, and most of what your dashboard shows you does not.

Who has this problem

You are a technical founder, so instrumenting things is the comfortable part. You added tracking early and thoroughly.

That is exactly why you now have too much of it. Collecting data is engineering, which you are good at. Deciding from data is judgement under uncertainty, which nobody taught you, and no quantity of charts converts one into the other.

If you have ever opened your analytics, looked at it for a minute, felt vaguely encouraged, and closed it without changing anything, this post is for you.

Why dashboards produce comfort instead of decisions

Three reasons, and they compound.

Vanity metrics only ever go up. Total signups, cumulative page views, stars, newsletter subscribers. They are running totals, and a running total cannot get worse. That makes them pleasant to check and useless as a warning system. A number that can only rise cannot tell you something is broken.

Rates get worse; totals never do. "1,400 signups" tells you nothing. "3% of visitors sign up, down from 5%" tells you something changed and roughly when. Only a rate can carry bad news, and bad news is the entire value of measurement.

Small numbers do not mean what they appear to. With 20 visitors you cannot measure a conversion rate. One extra signup moves it by five percentage points. Founders read that swing as a result, change the landing page, and have now made a decision based on one person's Tuesday.

There is a fourth reason, and it is the expensive one. Measuring feels like progress. An afternoon spent building a funnel report is an afternoon that did not involve talking to a stranger who might say no. It is the most comfortable form of avoidance available to a technical founder — which is why the builder's trap keeps working even on people who know about it.

The four metrics that earn their place

1. Conversations started

How many real prospects you spoke to or wrote to this week.

This is an input, and it is the only one on this list you fully control. Everything downstream is a fraction of it. When results are bad, this is the first number to check, and it is bad far more often than founders expect — because a week where you sent four emails and told yourself you had been doing outreach is a week that produced nothing.

Count it weekly, not cumulatively.

2. Reply rate

Replies divided by conversations started.

This is the first honest signal about your targeting and your message. Below roughly 2% on cold outreach, something is wrong and it is usually not the copy. Between 5% and 10%, it is working.

The diagnostic power is in what it separates: no replies points at the channel or the message; replies but no meetings points at the message or the offer. Two very different weeks of work.

3. Conversion to paid

Of the people who reached a real conversation, how many paid.

This one is brutal and worth measuring anyway. It is where an offer that is really a product description becomes visible. High reply rate and near-zero conversion is the classic signature of a product people find interesting and nobody needs.

4. Churn

Of the people who paid, how many are still here next month.

The only metric on this list that measures the product rather than the marketing. It is also the one that decides whether growing is worth doing at all — filling a leaking bucket faster is not strategy, and churn above roughly 10% monthly at this stage means the acquisition work is premature.

What to ignore for now

Not forever. For now.

None of these are fake. They are just downstream of the four above, and reading them first tells you how you feel rather than what to do.

The one piece of arithmetic to do

Before you spend anything on acquisition, work out these two numbers:

• Cost per customer — everything you spent on a channel, divided by customers it produced.
• What a customer is worth — monthly price × how many months they typically stay.

If you cannot state the first for a given channel, you are not ready to put money into it. That is not caution, it is the difference between buying growth and buying noise. It is also the gate that decides whether paid channels are available to you at all, which the first 100 customers problem usually resolves in favour of doing things that do not scale.

Why this framework holds up

The four metrics are ordered deliberately: each one is a fraction of the one above it. That makes the sequence a diagnostic rather than a scoreboard. Whichever conversion between two adjacent steps is worst is your bottleneck, and it is the only thing worth working on this week.

The Mark Platform tracks these four against your logged execution and tells you which one is the constraint, rather than showing you everything and leaving the interpretation to you. It also refuses to add conversion counts from two sources for the same channel — a signup seen by both your analytics and your ad platform is one signup, and summing them makes a failing campaign look profitable.

The honest limitation: none of this works if you are not logging execution consistently. A founder who logs two days out of twenty and then reads their reply rate is reading nothing. The metrics are downstream of the habit, not a substitute for it.

Key takeaway

If a number would not change what you do next week, stop looking at it. Four metrics, checked weekly, beat fourteen charts checked daily.

Frequently asked questions

When do the other metrics start mattering? Roughly when you have enough volume for them to be stable — a few hundred visitors a week, or twenty-plus customers. Before that they move for reasons unrelated to anything you did.

What is a good conversion rate? The wrong question this early. The useful one is whether yours is moving, and which adjacent step is worst. Benchmarks from companies with 100× your volume will mislead you.

How often should I look? Weekly. Daily checking on small numbers is how founders react to noise, and reacting to noise is worse than not looking.

What if all four numbers are bad? Start at the top. Conversations started is the input you control, and it is bad more often than anyone expects.

---

You know which four numbers to watch. The Mark Platform tracks them against what you actually did each week, names the bottleneck, and tells you the one thing to work on next. See what it measures and how, or start your marketing journey →