At some point the business stops being a question of whether anyone wants it and starts being a question of arithmetic. More people arrive than you can talk to individually. Some stay and some leave. Money comes in on a schedule.
This chapter is about that phase, and specifically about resisting the urge to do everything at once.
The loop that makes a business grow
Growth is not a list of tactics. It is a loop, run repeatedly:
- Look at the numbers. What is actually happening, not what you feel is happening.
- Name the constraint. The single thing most in the way right now.
- Make one change against it. A change big enough to move the number if you are right.
- Wait long enough to know. Then decide whether you were right.
- Go back to step one. The constraint has usually moved.
That is the whole method. Its power is entirely in the discipline of step two, because almost everyone skips it. Skipping it means doing five reasonable things at once, none of which touches the actual bottleneck, and then being confused when nothing improves.
Here is what the constraint usually is, in order, for a business that has its first customers:
- Not enough people arrive. You have a traffic problem. Fixing the product will not help.
- People arrive and leave immediately. You have a message problem. The page is not saying what the visitor needs to hear.
- People sign up and never come back. You have an activation problem. They did not get to the useful part.
- People use it and then stop paying. You have a retention problem, and it is the most expensive one to ignore.
Only one of these is your problem this month. Find out which.
Set up analytics you will actually read
Most analytics setups fail by measuring too much. A wall of charts nobody opens is worse than three numbers you check every Monday, because it feels like measurement while telling you nothing.
Set up four things and ignore the rest for now:
How many people arrive, and from where. You need to know whether last week's post did anything.
Where they drop off. The path from arriving to getting value has a few steps. Find the step where most people vanish. That step is your next month of work.
Whether people come back. One visit is curiosity. A second visit is interest. A fourth is a habit, and habits are what turn into revenue.
What people actually do. Which parts get used and which parts you built for nobody. Be prepared to be annoyed by this one.
Two rules to keep this honest. Look at real numbers, however small; twelve visitors is a real number and rounding it up in your head helps nobody. And check on a schedule, once a week, not obsessively. Daily checking of small numbers is a very effective way to feel bad without learning anything.
Know your numbers
There are six numbers that describe almost any subscription business. You do not need a finance background to use them, and knowing them changes how you make decisions.
Churn. The share of customers who leave in a given month. If you have a hundred customers and five leave in a month, that is five percent monthly churn.
Average customer lifetime. How long someone stays, on average. It is one divided by your churn rate. Five percent monthly churn means the average customer stays about twenty months.
Revenue per customer. Your monthly revenue divided by the number of customers paying. Useful mainly because it tells you what a customer is worth per month.
Lifetime value. Revenue per customer multiplied by average lifetime. If they pay fifty a month and stay twenty months, each customer is worth about a thousand.
Cost to acquire a customer. Everything you spent getting customers in a period, divided by how many you got. Include your own time if you are honest, and value it at what you would pay someone else.
The ratio between the last two. Lifetime value divided by acquisition cost. Under one, you lose money on every customer and growth makes it worse. Around three is generally considered healthy. Very high can mean you are underspending on growth rather than doing brilliantly.
There is a seventh worth watching: payback period, the number of months before a customer has repaid what it cost to acquire them. Lifetime value can look wonderful and still leave you short of cash if it takes fourteen months to arrive.
Two warnings. Early on these numbers are noisy, because a handful of customers leaving swings the percentage wildly. Use them as direction, not truth, until you are past a few dozen customers. And churn is the one to fix first, always. Growth on top of high churn is filling a bucket with a hole in it, and the bigger you get the faster it drains.
Support customers without hiring
Support is where solo businesses either build enormous loyalty or quietly drown.
The approach that works is to answer everything personally at first, then convert the repeats into something that answers itself.
Answer everything, fast, in the first few months. Speed matters more than polish. A quick honest reply from the person who built the thing is a competitive advantage that larger companies genuinely cannot buy.
Keep a tally. Every time you answer the same question a second time, write it down. That list is your entire support strategy.
Fix the top of the list in the product. A question asked repeatedly is usually a design problem wearing a costume. Confusing label, missing explanation, unclear next step. Fix it where it happens and the question stops.
Write the answer once, publicly. For the questions you cannot design away, write a short clear help page. Send the link, and keep the personal note on top of it.
Set expectations you can keep. "I reply within a day" that is always true beats "instant support" that is sometimes true.
The signal to watch for is a change in the shape of your questions. Early questions are about confusion, which is your problem to fix. Later questions become about capability, which means people have understood it and now want more. That shift is one of the better signs that the product is working.
Find the one thing holding you back
This is step two of the loop and it deserves its own section, because it is the step everyone skips.
At any moment, one thing limits your business more than everything else. Improving anything else produces almost nothing. The work is identifying it honestly, which is uncomfortable, because it is usually not the thing you enjoy working on.
A way to find it: walk the whole path a customer takes, and put a number on each step.
- People who see something about you.
- People who visit.
- People who understand what it is.
- People who sign up or get in touch.
- People who reach the useful part.
- People who pay.
- People who are still paying in three months.
Now look at the biggest drop between two adjacent steps. That is your constraint. Not the step with the lowest number, the step with the steepest fall.
Two rules that make this work. Only one constraint at a time, even when three look bad. And re-check it after every meaningful change, because fixing one bottleneck almost always creates the next one somewhere else.
If you are honest about this, you will occasionally discover that your constraint is that not enough people know you exist, and there is no product work that fixes it. That discovery is worth a month of building.
Run one experiment at a time
An experiment is a change plus a prediction. Without the prediction, it is just a change and you will interpret whatever happens as a success.
Write it as one sentence before you start: if we do X, then Y will improve, because Z.
Then respect three rules.
One at a time. Change two things and the result is uninterpretable. This is boring advice and it is the difference between learning and guessing.
Give it long enough. Long enough for a normal cycle of your business, usually two to four weeks. Small numbers move for random reasons and a good result on day two is almost always noise.
Write down what happened, including the failures. Especially the failures. A failed experiment that you remember is worth more than a successful one you cannot explain, because it stops you repeating it in five months when it seems like a fresh idea.
Expect most experiments to do nothing. That is the normal hit rate, even for people who are good at this. The value is not in each experiment, it is in the accumulating pile of things you now know are not the answer.
Decide when to expand
Eventually you will want to serve a second kind of customer, or add a significant new capability, or move upmarket. Some of those moves make the business, and some of them are how a working business gets broken.
Five questions to ask before any expansion:
Is demand already showing up? The best expansions are things customers keep asking for and occasionally try to hack together themselves. The worst are things you find intellectually interesting.
Does it help the numbers that matter? Will it reduce churn, increase what customers pay, lower your acquisition cost, or shorten payback? If it does none of those, it is a hobby with a business attached.
Does it strengthen the core or split it? A good expansion makes the main thing more valuable. A bad one creates a second product with a second set of support questions and a second audience to market to, run by the same one of you.
Can you test it cheaply first? A page describing it, a manual version done by hand for three customers, a waiting list. Almost anything can be tested before it is built, and almost nobody does.
Does it genuinely widen the market? Some expansions add work without adding reachable customers. Adding a feature the same hundred people wanted is fine, but it is not expansion, it is product work.
And one thing to hold on to as things grow: the reason your first customers chose you was probably specificity. Getting bigger by getting vaguer is the most common way a good small business becomes a forgettable medium one.
That is the series. Chapter one is where it starts, and starting is still the hard part.