Your First SaaS Ad Campaign: What to Do Before Paying

You put $300 through Meta. You got 6,000 impressions, 40 clicks, and no signups. The dashboard called it a successful campaign because it delivered what you paid for: impressions.

The usual conclusion is "ads do not work for us." The usual reality is that the campaign was optimising for the wrong thing, against an audience assembled from a guess, pointed at a page that was never going to convert cold traffic.

Ads amplify a funnel. They do not create one. And unlike almost everything else in marketing, ads spend real money while you find that out.

Who has this problem

You have a product with some organic traction — a few hundred signups, a handful of paying customers — and organic growth is slow. Ads look like the obvious accelerator, and the ad platforms are built to make starting extremely easy.

That is the trap. The distance between "create campaign" and "money leaving your account" is about ninety seconds, and every default in that flow is set for an advertiser with a working funnel and a tracking setup that has been correct for years.

If you have ever run a campaign and been unable to say afterwards whether it worked, this post is the checklist you were missing.

Why first campaigns fail on setup, not creative

Founders assume the problem was the image or the headline. It is almost never the creative. It is one of three setup failures, and all three are invisible from the ad platform's own dashboard.

You are optimising for a conversion the platform has never seen. A Meta pixel can be installed, firing page views, and looking perfectly healthy in the browser — while never once having received a purchase or signup event. In that state a conversion campaign has nothing to learn from, so it never leaves the learning phase and spends the budget on whoever is cheapest to show an ad to. You cannot detect this by inspecting your website. It only shows up if something queries the pixel's actual event history.

Your targeting is a guess dressed as a strategy. Interest targeting built from what you imagine your customer likes produces an audience nobody has validated. If you have not written down who the buyer is and what they are trying to fix, the campaign is an expensive way to find out.

Nothing forced you to look before it spent. Campaigns go live immediately. There is no natural checkpoint between "this looks about right" and "this is now spending £40 a day", and the first review usually happens after the first invoice.

The checklist that has to pass before you spend

Step 1: Do not run ads without an offer and a persona

Cold traffic converts on a specific promise to a specific person. If you cannot fill in "I help [who] solve [what] by [how]", the ad has nothing to say, and no targeting will rescue it.

This is not a delay tactic — it is the cheapest step here, and it is why defining your ideal customer profile comes before any paid channel. The Mark Platform refuses to build a campaign from an incomplete journey and names the missing fields instead of generating something plausible from gaps.

Step 2: Verify the pixel receives a conversion event

Not that it is installed. That it has received a conversion. There are four states worth distinguishing: no pixel, installed but silent, page views only, and genuinely tracking conversions. Only the fourth supports a conversion campaign.

Check this before your first campaign and after any deploy that touched the signup flow.

Step 3: Preview the entire campaign before it exists

Targeting, ad copy, budget, estimated spend, destination — all of it, rendered, while touching no ad platform and creating nothing.

The Mark Platform makes this mandatory: the preview returns a hash, and the create step will not run without it. If your journey data changed between previewing and creating, the hash is refused and you preview again — so you can never approve one campaign and ship a different one.

Step 4: Create everything paused

Every campaign, ad set and ad is created in a paused state and spends nothing until you explicitly resume it. This gives you the checkpoint the ad platforms do not: the campaign exists, you can inspect it in Meta's or Google's own interface, and it costs nothing while you do.

It also means a half-finished creation is harmless. If any step fails, everything already created is deleted rather than left as an orphaned campaign you discover on next month's invoice.

Step 5: Set the spend cap and the CPA ceiling before you resume

Decide two numbers in advance: the most you will spend before stopping, and the cost per acquisition above which this channel is not worth it. Write them down while you are calm, because you will not be objective on day three.

Your CPA ceiling comes from what a customer is actually worth. If you are charging $10 a month with 30% annual churn, a $90 CPA is not an aggressive bet, it is a loss you have chosen.

Step 6: Declare special ad categories yourself

Credit, employment, housing, and social or political issues carry mandatory restrictions on targeting. Never let a tool infer this from your campaign text — the inference will sometimes be wrong, and the consequence is a disabled ad account rather than a warning. Declare it explicitly or declare that it does not apply.

Why this framework holds up

Every step above is a checkpoint that costs nothing and catches something that costs money. That is a deliberately unglamorous design: preview before create, paused before live, guardrails before resume, human declaration before a restricted category.

The Mark Platform also refuses a conversion objective when no pixel is tracking conversions, rather than quietly downgrading the campaign to traffic. A downgrade would have produced a campaign that runs, spends, delivers clicks, and teaches you nothing — which is worse than a refusal, because it looks like it worked.

Where this will not help: ads make an existing funnel bigger, including its leaks. If 90% of signups never activate, paid traffic buys you more people to lose. Run a funnel drop-off analysis first and fix the worst step — that is free, and it changes the return on every pound you spend afterwards.

Key takeaway

Verify the pixel actually receives conversions, preview the whole campaign before it exists, create it paused, and decide your spend cap and CPA ceiling before you resume. Creative is the last thing to optimise, not the first.

Frequently asked questions

How much should I budget for a first test? Enough to reach statistical meaning on your funnel, which for most small products means a few hundred over two to three weeks rather than $20 a day indefinitely. Underfunding a test produces a result you cannot act on.

Meta or Google first? Google if people already search for the problem you solve — that is existing demand and it converts faster. Meta if they do not know the category exists and have to be shown it.

Should I run ads before I have paying customers? Usually not. Ads are the most expensive way to learn who your customer is. Learn that through interviews and organic channels, then use ads to reach more of a person you have already validated.

My ads got clicks but no signups. What broke? Almost always the landing page or the match between ad and page. If the ad promises one thing and the page says another, the click was wasted at the moment it landed.

How long before I stop a losing campaign? Once you have hit your predetermined spend cap. That is what deciding it in advance was for.

---

You have the checklist. The Mark Platform previews the campaign before it exists, creates everything paused, and refuses to build one from an incomplete journey. See what it costs, or start your marketing journey →