When should a startup start running paid ads?

A startup should start running paid ads only after it has confirmed, through direct manual effort, that it knows exactly who wants its product and why. Before that threshold, ad spend masks the signal you need most: whether your product genuinely solves a real problem. Most early-stage startups hit paid acquisition too soon and mistake traffic for traction.

Why paid ads are the wrong first move

Paid advertising is fundamentally an amplification tool. It takes a working message, a proven offer, and a converting landing page, and puts them in front of more people faster. What it cannot do is discover whether your product is worth buying in the first place. When you run ads before you have product-market fit, you are paying to learn the wrong things — which ad creative performs slightly better — instead of the right things, which is whether your core value proposition actually resonates with real customers.

Paul Graham's argument in 'Do Things That Don't Scale' is that early founders should be doing the opposite of automated, scalable acquisition: getting out in person, recruiting users one by one, and learning things from direct contact that no analytics dashboard can surface. Airbnb's survival in its earliest days came down to roughly 30 days of founders physically visiting users and improving the experience by hand. That kind of intimate, unscalable work is what produces the insight that eventually makes paid acquisition efficient. Without it, you are bidding on keywords for a product you don't yet fully understand.

The practical implication is that if you are considering paid ads and you cannot clearly articulate the exact sentence that makes a specific type of person immediately want to try your product, you are not ready. That sentence — the real one, not the polished marketing version — comes from dozens or hundreds of real conversations, not from A/B testing ad copy.

The real prerequisite: manual distribution first

Before paid ads make any economic sense, you need to have acquired your first meaningful cohort of users through channels that require your personal involvement: cold outreach, communities, referrals, partnerships, content, and direct sales. This is not a temporary inconvenience to push through on the way to 'real' marketing. It is the primary research phase. Each manually acquired user teaches you something about why someone chose you, what they almost didn't, and what language they use to describe the problem you solve. That language is the raw material for every ad you will eventually write.

For B2B startups especially, the right move early on is often to treat one or two early customers almost like consulting clients — understanding their workflow at a granular level before you try to systematize acquisition. Once you have served a small number of customers well enough that they refer others without being asked, you have a signal that something is working. That referral behavior is meaningful evidence. Paid ads then become a way to find more people who look like your already-happy customers, which is a very different — and much cheaper — way to run ad campaigns than guessing cold.

A useful internal test: if your best current customers tried to describe your product to a colleague, what would they say? If you know the answer to that question with confidence, because you have heard it happen, you are close to being ready for paid acquisition. If you are guessing, you are not.

The specific signals that say you are ready

There are a handful of concrete indicators that suggest paid ads will generate return rather than just data. First, your conversion rate from landing page visitor to signup or purchase is high enough, based on organic or referral traffic, that you can calculate a rough payback period. If you do not know your conversion rate from non-paid traffic, you have no baseline to set paid budgets against. Second, you have identified at least one customer segment where you consistently win — where the same type of person, described with enough specificity that you could find them on an ad platform, keeps converting and staying. Third, your retention or repeat usage is strong enough that acquired customers actually generate the LTV your unit economics require.

If all three of those are true, even at a small scale, paid ads can work. If one or more is missing, you are not running an acquisition problem — you are running a product or positioning problem, and no amount of ad spend will fix it. Founders sometimes hope that volume will surface the answer, that if they just get enough people in the funnel something will stick. In practice, volume before fit mostly generates noise, churn, and a depleted bank account.

A secondary indicator worth watching: if your cost per acquisition on small, manually targeted campaigns — even just boosted posts or narrow search terms — is already within a range that works economically, you have a green light to scale. If CPA is far outside any defensible unit economics even on tightly controlled tests, you need to fix something upstream before putting more money in.

How to run your first paid experiments without wasting money

When you do start, treat the first paid campaigns as experiments with a fixed learning budget, not as growth investments. The goal is not to acquire customers at this stage — it is to confirm that the message and audience you identified through manual work holds up at the small scale of paid distribution. Spend a small, defined amount per week, isolate one variable at a time (audience or creative, not both), and measure against a single conversion event that is as close to 'user gets value' as possible, not just clicks or impressions.

Start with the channel where your target customer's intent is clearest. For most B2B products, that means search ads targeting specific problem-aware queries, not broad awareness. For consumer products, it often means a narrow interest-based audience on a social platform, focused on people who already know they have the problem you solve. Broad awareness campaigns are almost never the right first paid move for an early startup — they are expensive, slow to teach you anything, and they optimize for metrics that do not correlate with actual business value.

Set a clear stopping rule before you start: if you spend X amount and conversion rate is below Y, you stop and go back to talking to users. This rule protects you from the sunk-cost trap of continuing to fund campaigns that are not working because you have already invested in them. The learning from a failed paid experiment is still valuable — it tells you that your message, your audience targeting, or your product itself needs more work — but only if you cut it off while you still have enough runway to act on what you learned.

“Almost all startups are fragile initially. And that's one of the biggest things inexperienced founders and investors get wrong about them.”

— Paul Graham, source

The one thing to do

Before running a single paid ad, confirm through direct manual acquisition that you know exactly who wants your product and why — then use ads only to find more people who match that profile.

Frequently asked questions

Can paid ads help a startup find product-market fit?

Rarely. Paid ads can generate traffic but they produce weak signal on whether your product actually solves a real problem. Direct manual outreach and personal conversations generate the insight that eventually makes paid ads efficient. Use ads to confirm fit, not discover it.

What is the minimum viable test for paid ads?

Pick one narrow audience, one specific message, one conversion event, and a fixed maximum spend. If you cannot calculate a rough payback period from the results, your baseline metrics are not ready for paid acquisition yet.

Does the right timing differ between B2B and consumer startups?

Yes, modestly. B2B startups often benefit from delaying paid ads longer because direct outreach and content can drive higher-quality pipeline at lower cost. Consumer startups may test paid channels slightly earlier, but still only after confirming retention in a manually acquired cohort.

What if a competitor is running paid ads heavily — should we match them?

No. Competitor ad spend is not a signal that the strategy is working for them, and it is certainly not a signal that it will work for you at your current stage. Execution quality and product-market fit matter far more than whether you are visible on the same channels.

Sources

More playbook answers · Growth Prophet home