The short answer: SaaS customer acquisition changes with stage. From zero to ten customers, acquire by hand through your network, communities and direct outreach. From ten to a hundred, double down on whichever of those produced paying users and add referrals and outbound at volume. Past a hundred, content, search, partnerships and paid ads start to pay back. Measure what each customer costs you in time and money before scaling any channel.
Most advice on acquiring SaaS customers is written by companies with a marketing team, a budget and an audience. If you have none of those, following it produces a lot of activity and very few customers.
The short answer is that customer acquisition is not one strategy. It is a different job at zero customers, at ten and at a hundred, and the most common mistake is using the playbook for a later stage too early.
How do I acquire customers for my SaaS?
SaaS customer acquisition is the repeatable process of turning people who have a problem into paying subscribers, at a cost lower than what they will pay you over their lifetime. The word that matters is repeatable. A single good week is not acquisition. A method that produces roughly the same result each time you run it is.
That definition gives you a test for every channel you try: did it produce customers, and would it produce them again if you did the same thing next month? Most early channels fail the second question, which is fine as long as you notice.
The three stages, and what works in each
| Stage | Main goal | Channels that work | Channels to postpone |
|---|---|---|---|
| 0 to 10 customers | Learn who buys and why | Your network, communities you belong to, direct one-to-one outreach | Paid ads, SEO as a primary channel, launches |
| 10 to 100 customers | Find one channel that repeats | Outbound at volume, referrals, focused communities, a launch with real users | Broad brand campaigns, conferences |
| 100+ customers | Lower cost per customer | Content and search, partnerships, paid ads, product-led loops | Doing everything by hand |
Read the table by row, not by column. The channels in the right-hand column are not bad. They are expensive at the wrong stage, either in money or in the months they take to produce anything.
The boundaries are approximate. Ten and a hundred are useful markers, not rules. What actually moves you from one stage to the next is a change in what you know: first you learn who buys, then you learn which channel reaches them reliably, then you learn how to make that channel cheaper. If you skip a stage, you are trying to make cheaper something you have not yet made work.
Zero to ten: acquire by hand
At this stage you are not optimising a funnel, because you do not have one. You are finding out who buys. Every customer should come from a conversation you were part of, so you hear the objections and the exact words people use.
Where the first ten come from
- People who already know you. Former colleagues, clients, friends who run small businesses. They will take a call, and they will tell you the truth.
- Communities where your buyer is already active. Only if you participate as a member, not as a vendor.
- Direct outreach to people who describe the problem in public. Forum posts, GitHub issues, reviews of competitors.
This is covered step by step in how to get your first customers with no sales experience. The point here is that it is meant to be manual. Automating it before you know what works only spreads an untested message faster.
Keep a simple log of every conversation: who it was, where you found them, what they said the problem cost them, and whether they paid. Ten rows of that log will tell you more about your market than any analytics dashboard, because at this volume the numbers in a dashboard are noise and the words in a conversation are signal.
Expect most of these first customers to need a nudge. Few people sign up and pay on their own for an unknown product from an unknown founder. Offering to set it up with them, or to migrate their data, is often what turns interest into a first payment, and it shows you exactly where onboarding is too hard.
Ten to a hundred: find the channel that repeats
By ten customers you should know which source produced the paying ones. Maybe most came from one community. Maybe direct outreach to agency owners converted and outreach to developers did not. That pattern is the most valuable thing you own, and the job now is to run it again at higher volume.
Outbound at volume
If one-to-one outreach worked, it can scale, but only if each message still reads as written for one person. Volume without relevance is how sending domains get burned. Before you scale, read cold email deliverability so the channel that works does not quietly stop working.
Referrals
Happy early customers know people like them. Most will not refer anyone unless asked, and they need something easy to forward: one sentence and a link. Ask right after a customer gets a result, not at signup, when they have not seen anything yet.
A launch, now that you have something to show
Product Hunt, Hacker News and similar launches work far better with ten real users who can comment honestly than with zero. Treat the launch as a spike that feeds your repeatable channel, not as the channel itself.
Resist the urge to add channels here. Adding a second channel before the first one is working halves your attention and makes both look weak. Run one channel for long enough to see a result you trust, usually a month of consistent effort, then decide whether to keep going, fix it or switch.
How to work out what a customer costs you
Customer acquisition cost is everything you spent to win customers in a period, divided by the number of customers you won. Founders usually count tools and ads and forget their own time, which at the early stage is the largest cost by far.
A simple worked example, with made-up round numbers so the arithmetic is clear: you spend 20 hours a month on outreach and $100 on tools, and you win 4 customers. If you value your time at $50 an hour, the month cost $1,100, so each customer cost $275. If your plan is $40 a month, a customer needs to stay about seven months before they have paid back what it took to win them.
That number tells you two things. Whether the channel is worth scaling, and which lever matters more for you right now: lowering the cost of each customer, or keeping them longer. For a broader view of what each tool in a small stack actually costs, see the solo founder GTM stack, priced.
A hundred and beyond: lower the cost per customer
With a hundred customers you have enough data to see which channel has the lowest cost and the best retention. This is when slower, compounding channels earn their place, because they lower the cost of every future customer rather than adding customers one by one.
- Content and search. Articles that answer the questions your buyers type into Google or ask an AI assistant. They take months to rank, so start before you need them.
- Partnerships. Tools your buyers already use, agencies that serve them, communities that would recommend you.
- Paid ads. Useful once you know your conversion rate and what a customer is worth, so you can bid with confidence instead of guessing.
- Product loops. Features that put your product in front of new people as a side effect of normal use, like shared documents or public pages.
This is also the stage where retention becomes an acquisition lever. A customer who stays twice as long makes every acquisition channel twice as affordable, because the same cost to win them is spread over more months of revenue. Founders who focus only on new customers at this stage often find that fixing why people leave does more for growth than any new channel.
Content deserves a specific note, because answer engines now matter as much as classic search results. People increasingly ask an AI assistant how to solve their problem, and the assistant quotes pages that answer the question directly and honestly. Writing that kind of page is the same work as writing for search, and it compounds the same way.
Where this framework breaks
The stages are a simplification, and they fit some products badly.
- Very low-priced products cannot afford one-to-one acquisition at any stage. At $5 a month, you need content, communities and product loops from the start, and the manual phase is mainly for learning.
- Products sold to large companies spend far longer in the first stage, because each customer involves several people and a long decision.
- Products with a strong existing audience can skip the first stage. If you already have a newsletter of ten thousand of your buyers, use it.
It also assumes the product solves a real problem. If careful outreach to the right people produces interest but no one pays, acquisition is not the bottleneck. Distribution problem or product problem helps you tell the two apart.
Where an AI sales agent fits
The middle stage is the one most builders get stuck in. They know what worked for the first ten, but repeating it means hours of research, writing, follow-ups and scheduling every week, which competes directly with building. That repetitive middle is what Operater's Sales agent is designed to run: it finds people who match the buyer you have identified, writes to each one personally, handles replies and books the conversation. It is in closed beta, and it is only as good as the buyer definition you give it, so the manual first stage is still yours to do.
Key takeaways
- Customer acquisition is stage-dependent: the channel that finds ten customers rarely finds a thousand.
- Below ten customers, every acquisition should be a conversation you can learn from.
- Customer acquisition cost includes your own hours, which is the cost founders most often ignore.
- A channel is worth scaling only when it works twice in a row with roughly the same effort.
- Search, paid ads and partnerships compound later, so start them early but do not depend on them early.