Why metrics mislead
After a launch you will have many numbers and little time. The risk is to look at the largest ones because they feel good. Page views and visitor counts are large and flattering. They are also the least informative.
A useful metric is one that changes what you do. If you cannot say what you would do differently if it moved, it is probably decoration.
The stages
Think of your product as a series of steps. A person moves through them, or does not.
Visitors. The number of people who reached the page. It measures reach. It tells you nothing about interest, and it can be inflated by bots and curiosity.
Signups. People who gave an email or created an account. It measures interest. It can be inflated by incentives, giveaways, or low commitment offers.
Activation. People who did the thing that makes the product useful, such as completing setup or getting a first result. It measures value received. This is the first metric that shows the product works. See user activation for a new product.
Return. People who come back after the first use. It measures habit. A product that no one returns to is a product that did not stick.
Payment. People who paid. It measures demand in its clearest form.
What each one can hide
A rising visitor count can hide that none of them signed up. A rising signup count can hide that none of them activated. High activation can hide that nobody comes back. And a few payments can hide that all of them came from your friends.
Always read a metric together with the one after it.
Ratios matter more than totals
Compute the share who move from each step to the next. Visitors to signups, signups to activation, activation to return, return to payment. The weakest step is where to work. The method is similar to the one in waitlist conversion rates, with real numbers.
Split by source and by user type
An average hides important differences. Your best users may come from one place and share one trait. Look at each group separately.
What we do not give you
We do not give you target numbers. Published benchmarks for these measures come from different products, different audiences, and different definitions, and we have not found ones we can source and date properly. Compare yourself with yourself, week to week.
Small samples
In the first weeks, your totals will be small. A few users can swing a percentage a long way. Describe what you see, avoid sweeping conclusions, and keep collecting.
A one page dashboard
You do not need a complicated tool. A weekly table with these columns covers the essentials: week, visitors, signups, activated, returned, paid, and a notes column.
Fill it in each week and add a second table by source. After a month you will see where the largest drops are. Write one sentence below the table: “The biggest drop is between X and Y, and I think it is because Z.” Check that sentence the following week.
Metrics that sound useful and often are not
- Total page views. Includes bots, repeat visits and your own testing.
- Followers. Measures reach on a platform, not interest in the product.
- Time on page. Can mean engagement or confusion.
- Downloads or signups without use. Show interest, not value.
Use them for context, not for decisions.
Metrics that deserve attention
- The share of signups who complete the key action
- The share of activated users who return within a week
- The number of people who pay without being asked twice
- The number of people who reply when you write to them
Each is small and can be counted by hand at the start. Do not wait for a perfect tool.
Definitions matter
Write down what you mean by each term. If “active” means logged in for you and used the main feature for your colleague, your numbers will drift apart. A definition written once and kept stable lets you compare weeks honestly.
Where to go next
If the numbers look weak at the first step, read why the launch spike does not convert. If they look weak after signup, see when nobody buys. For the whole picture, return to what happens after you launch.
