Paid Ads: a Tax on Hope for Start-Ups

There's a post that appears on Indie Hackers every few weeks. A founder spent $1,075 on Reddit ads. Zero signups. Zero customers. Zero leads. They describe the targeting, the creative, the landing page, then ask what they got wrong. They didn't get anything wrong tactically. They got the sequencing wrong. They tried to buy distribution before they'd built the conditions that make distribution purchasable. That's not a campaign failure. It's a GTM sequencing error.

The average customer acquisition cost for B2B SaaS sits at around $1,200 across all channels in 2025, and it's been rising roughly 14% a year. That's the blended number. Paid channels run higher.

What this means in practice: if you're under $2M ARR, every dollar you spend on paid ads is almost certainly being spent before the conditions that would make paid work actually exist. You're not buying customers. You're buying a flat dashboard and a feeling of forward motion.

This piece explains exactly what those upstream conditions are, how to know if you have them, and what to do with your time and budget until you do.

What Paid Ads Actually Require Before They Return Anything

Paid acquisition is a multiplier. If the thing you're multiplying isn't working yet, paid makes it more not-working, faster, and at a cost-per-click you're paying for the privilege. Before a paid channel can return anything, you need four things in place. Most early-stage founders have none of them.

1. A landing page with a measured conversion rate — Not guessed. Not inferred from three sessions in Hotjar. Measured, meaning you've run enough traffic through it to know what percentage of strangers convert, and why. Without this, you're pouring money into a vessel with an unknown hole in it.
2. A message a stranger can repeat back after one glance — Not a value prop written in a Google Doc. Not a tagline your co-founder agreed sounded good. A line that lands in the reader's head in under three seconds, which you know works because an actual stranger has repeated it back to you unprompted.
3. Unit economics that survive the auction — Paid platforms are auctions. The price of a click in your category is set by every other company bidding for the same audience, including the well-funded ones with optimized funnels and CAC data from 50,000 customers. If your LTV assumptions come from an industry playbook rather than your own retention data, you don't know whether you can afford to play.
4. Enough volume to generate learnable data — A hundred clicks is not a sample. Two hundred clicks is barely a sample. At typical B2B SaaS conversion rates, you need somewhere between 500 and 1,000 clicks to start drawing conclusions about a single ad variant. Most early-stage paid experiments run out of budget before they've generated anything except noise.

Take a founder at month 18, $3K MRR, building a project management tool for engineering teams. They have a landing page they built over a weekend, a headline they split-tested between two variants over two weeks at low traffic, LTV estimates from a benchmarks report, and a $1,200 monthly ad budget. That budget buys them maybe 300 to 400 clicks at B2B SaaS CPCs. At a 2% conversion rate — optimistic for a page that hasn't been properly tested — that's six to eight leads. That's not data. That's a guess with a receipt.

The Sequencing Error: Why Founders Run Ads Too Early

Paid ads feel like action. You write a headline, you set a budget, you watch the impressions counter tick up. There's a dashboard. The dashboard has numbers. The numbers change.

Pre-PMF distribution work — the kind that actually teaches you something — doesn't come with a dashboard. It's ten conversations with prospective customers you had to go and find. It's spending forty minutes in a Slack community where your ICP lives, reading what they complain about. It's an email to twenty people who fit the profile, written one at a time, asking a question you actually want the answer to.

That work is slower. It doesn't scale. It doesn't feel like a startup is supposed to feel. And for a technical founder whose entire professional identity is built around building systems that scale, doing manual, unscalable things feels like regression. This is the trap. The whole premise of the paid experiment is 'let's skip to the part where this works at scale.' But scale doesn't work until the unit does. The unit is the conversation. You don't get to skip it by spending money.

Here's what the sequencing actually looks like:

A founder at month 8, $0 MRR, building a developer tooling product, spent six weeks doing exactly this: 23 conversations, three community posts, two cold email sequences to 40 people each. At the end of it, they had three phrases that three or more people had used without prompting. One of those phrases became their headline. Their landing page, running on organic SEO traffic, converted at 4.2%. That's when paid became a viable next step. Not month one.

The Fifteen Levers and Where Paid Sits in the Chain

There are fifteen to twenty levers inside the commercial layer of an early-stage company: positioning, pricing, messaging, ICP definition, distribution channel selection, funnel shape, sales motion fit, team structure. Paid acquisition is one of these levers. It sits near the end of the chain. Not the beginning.

The levers at the beginning — the ones that determine whether paid can ever work — are almost always untouched in pre-PMF startups. They were built for the pitch deck and left alone. Here's the order they need to be addressed:

• ICP definition: Who specifically is this for? Not 'SMBs' or 'dev teams'. A specific role, in a specific context, with a specific problem that has a cost attached to it.
• Positioning: What is this, for that person, compared to what they're doing now? One sentence, no hedging.
• Messaging: What do you say to that person that makes them stop scrolling? Derived from the conversations, not from a brainstorming session.
• Distribution channel selection: Where does that person actually spend time, and what does reaching them cost?
• Conversion mechanics: What does a stranger do when they land, and what percentage do it?

Paid is the sixth step, not the first. When founders skip to step six, they're not being aggressive about growth. They're being expensive about avoidance.

The agent tooling now available to run paid experiments in an afternoon is genuinely impressive. The problem is that it lets founders skip the upstream steps faster. The speed is the danger.

What to Do With the $1,075 Instead

This is the actionable version. If you have a budget and you're under $2M ARR, here's where it goes before it touches a paid channel.

Week 1: Run ten conversations

Message ten people who fit your ICP. LinkedIn, community Slack, email, mutual introductions, whatever it takes. Don't pitch. Ask: what are you using to solve this problem right now, and what does it cost you when that doesn't work? Record what they say. Write it down verbatim. Cost: four hours. No spend required.

Week 2: Extract the language

Look at your notes. Find the phrases that recurred. Find the problems they named in their own words. Find the objections. Build a message from their words, not yours. Test it in a cold email sequence to twenty more people. Measure the reply rate. Cost: two hours plus any email tooling you already have.

Week 3: Rebuild the landing page around what you heard

The headline is the phrase that came back three times. The subheading names the cost of not solving the problem. The CTA is specific. Run this version against whatever traffic you currently have. Cost: a day of work.

Week 4: Measure

Is the page converting? Is the reply rate on the cold emails above 15%? Are the conversations turning into trials or paid conversations? If yes: you now have something worth amplifying. If no: you have more signal to work with. Either outcome is worth more than $1,075 in Reddit ad spend.

The conditions that make paid work are built from conversations and iteration, not from ad budget. Save the money until you know what you're saying, who you're saying it to, and that it's landing.

Where to Start This Week

If you're sitting under $2M ARR with a paid budget and a flat dashboard, the next step isn't a better creative or tighter targeting. It's ten conversations with people who look like your best prospective customer.

Write down what they say. Look for the phrases that come back. Build the message from those phrases. Test it in zero-cost channels first. When a version of the message works, that line is the paid ad.

The commercial layer of an early-stage company has fifteen levers. Tincture's free self-assessment maps where yours are set and which ones are doing nothing. It takes twelve minutes and tells you whether paid is even in the right place in your sequence yet.