November 18, 2026
PPC for SaaS fails for a reason most dashboards hide: you are paying for clicks and judging yourself on a signup, while the number that pays your salary is a customer who stays past month three. The gap between those three events is where budgets die.
This article gives you a working model. What to count as a conversion, how much you need to spend before a test means anything, how to read ROAS and CPA without fooling yourself, and when to kill a campaign instead of "optimizing" it one more week.
If you are a founder spending your own money, read the budget floors first. Those are the numbers that decide whether any of the rest matters.
Start by picking the conversion you optimize to
A conversion is the single event you tell the ad platform to chase. Everything the algorithm does bends toward producing more of that event, so choosing the wrong one quietly wastes most of your spend.
The honest answer for most SaaS is that your best money event (a paying, retained customer) happens too late and too rarely for the platform to learn from. So you pick a proxy. The trade-off runs along one axis:
| Optimization event | Signal volume | Correlation with revenue |
|---|---|---|
| Any lead or email | High | Weak |
| Qualified trial or demo request | Medium | Medium |
| Activated trial (used the product) | Lower | Strong |
| Paid conversion | Lowest | Exact |
Pick the latest event in that table that still fires at least 30 to 50 times a month per campaign. Below that volume, the platform's optimization is guessing, and you are better off feeding it a higher-volume event upstream and filtering quality yourself.
If you run demos or sales-assisted deals, send the back-end event (the closed deal, or the sales-qualified stage) back to the platform through offline conversion import. That is what lets the algorithm learn from revenue instead of form fills. The exact import path and label change often, so check the platform's current documentation for the setting rather than trusting a screenshot from last year.
Set a spend floor before you call anything a test
A test that cannot reach statistical honesty is just an expense. Before you launch, work out whether your budget can produce enough conversions to read a result inside your patience window.
Rough working rule: you want roughly 30 or more conversions on the optimization event before you trust the direction of a campaign, and you want them inside two to four weeks so the market has not shifted under you.
Back into the spend from there:
- Estimate your target cost per conversion. If you have no history, take your rough target CPA and double it for the learning period.
- Multiply by 30. That is your minimum to see a signal.
- If your optimization event is a demo at, say, an estimated $150 each, you need roughly $4, 500 to read one campaign, not $500 across five.
The most common founder mistake is splitting a small budget across many campaigns and audiences. You end up with ten things, none of which ever gathered enough data to tell you anything. Concentrate. One campaign, one clear event, enough budget to learn.
Match the channel to intent, not to hype
Different paid channels sell you different things. Search sells intent. Social sells interruption. You pay for each one differently and you read each one differently.
Search: capturing demand that already exists
Someone typing your category or a competitor's name is raising their hand. That is the most efficient SaaS spend available, and also the most competitive. Expect high cost per click on bottom-funnel terms and defend your margin by being ruthless about which queries you actually pay for. This is where tight negative keyword lists and ongoing query mining earn their keep. A structured approach to PPC Competitor Analysis tells you which rival terms are worth bidding on and which just drain budget.
Paid social: creating demand that is not searching yet
On social you interrupt people who were not looking for you, so your creative does the work the query would have done on search. Expect longer payback and a messier attribution picture. Treat it as pipeline generation, not last-click revenue, and judge it over a 28-day window rather than a 7-day one.
For the full build across both, including how to sequence them as budget grows, the B2B SaaS Marketing: A Paid Media Playbook for Growth walks through the channel mix. If you are deciding whether paid is even the right place for the next dollar against organic, SEO vs PPC: How to choose lays out the comparison.
Read ROAS and CPA with the window attached
A ROAS number without an attribution window is not a claim, it is a vibe. A 7-day-click ROAS and a 28-day-click ROAS describe different realities, and SaaS sales cycles make the gap enormous.
Rules that keep the numbers honest:
- Always state the window. "3x ROAS" means nothing. "3x on a 28-day click window" is a claim you can defend.
- Pick one window and hold it. If you switch from 7-day to 28-day halfway through a quarter, every trend line you drew is now fiction.
- For subscription products, ROAS on first payment understates reality and ROAS on projected lifetime value overstates it. Report both, and be clear which one you are looking at when you decide to scale.
- Platform-reported conversions and your own database will disagree. They always do. Reconcile them monthly and trust your database for the money decisions.
The deeper problem is that paid platforms count a conversion inside their own window and take credit generously. Your SaaS marketing funnel is the only place you see the whole path, so build the funnel view before you let any dashboard tell you a channel is working.
Know the failure modes before they cost you
Most wasted SaaS ad spend comes from a short list of repeatable mistakes. Learn to spot them early.
- Optimizing to a cheap, abundant event (free email signups) that does not convert to revenue. You scale the campaign, cost per signup drops, and revenue does not move.
- Scaling before the payback period has elapsed. You cannot know your true CAC on a 30-day cohort if customers churn in month two. Wait for the cohort to mature.
- Letting broad match and loose audiences spend on the wrong people. Check your search terms and placement reports weekly, not quarterly.
- Judging social on last click. It rarely wins last click and often starts the journey. Kill it on last click and you kill your cheapest top-of-funnel source.
The discipline that separates accounts that scale from accounts that stall is a weekly review tied to cohort data, not a dashboard refresh. A repeatable PPC Account Management: A Framework matters more here than any single clever bid tactic.
When to kill a campaign
Kill it when it has had enough budget to prove itself and the mature cohort still loses money. The trap is killing too early on noise, or too late out of hope.
Decision rule: if a campaign has spent your 30-conversion floor, the cohort has aged past your payback window, and blended CAC still sits above what you can afford against lifetime value, stop. Do not give it one more week. One more week is how a dead campaign eats a quarter.
If blended CAC is drifting up across the whole account while individual campaigns all look fine, that is channel saturation or audience overlap, not a campaign problem. That calls for a structural rethink of the account, which is where a specialist review pays off. Our PPC Agency work usually starts by rebuilding the measurement before touching a single bid.
Frequently asked questions
What to do next
Pick one conversion event, confirm your budget can hit roughly 30 of them a month, and set a single attribution window you will not change this quarter. That alone fixes most of what is broken in a struggling account.
If you want a second set of eyes on the measurement before you spend more, start a conversation with our B2B PPC: Build Campaigns team and have your funnel numbers ready.

