Google Ads for SaaS: capture demand you already earned

Search ads are a spreadsheet before they are a channel. Work out your LTV, set a blended CAC ceiling at roughly half of it, and pick a hard kill line just above the ceiling. Every operating decision later (scale, hold, cut) is made against those two numbers, never against how a campaign feels.

The channel verdict for most early SaaS is workable but tight: roughly break-even cash over the first months against expected LTV. You run it for the compounding keyword and copy knowledge, then scale the parts that beat the ceiling.

Start with exactly two Search-only campaigns and give the higher-intent one its own budget so the other cannot starve it:

1. 1Competitor and alternative terms: people typing a rival's name or [rival] alternative are already shopping for your category. Highest intent, cheapest entry bids, usually the workhorse.
2. 2Category generics: what your buyer types when they do not know any brand yet. Bigger volume, colder intent, so cap its bids lower.
3. 3No brand campaign while organic already owns your name, no Display, no Performance Max, search partners OFF. Every one of those spends your learning budget on lower-intent inventory.

Send clicks to public pages that already convert organically, never to a login-walled app screen. The signup flow is the conversion path, not the landing page.

Broad match hands your budget to Google's guesses. Use exact [brackets] and phrase quotes only, and attach a shared negative list before the first impression:

Responsive search ads want up to 15 headlines; fill them with the reasons buyers actually choose you (what it does, the price, the objection killers), not adjectives. Two rules save you from expensive lessons:

Google optimizes toward whatever you report, so a broken conversion event does not just miscount, it steers the money wrong. The classic failure: the purchase event lives on a checkout success page, but your real buyers convert through an in-app paywall that never visits that page. Zero conversions reported, four real customers, and the account learning from silence.

1. 1Map every path a buyer can take to a paid subscription, and fire the purchase event on all of them, deduped by transaction id
2. 2Report the cash collected today as the conversion value (an annual plan reports the annual charge), so bidding optimizes toward money, not signups
3. 3Add signup as a secondary observation-only conversion; never let the account bid on it
4. 4Keep auto-tagging on and store the click id and UTMs on the user record at signup; that is what lets you compute true CAC from your own database
5. 5Missed conversions are recoverable: export the click ids from your database and use offline conversion import

Verify with ground truth, not the pixel: query your own database for ad-attributed signups and customers in the first 48 hours. If the Ads UI and your database disagree, your database is right and the tracking needs fixing.

The account is a weekly habit, not a daily hobby. The cadence that keeps it honest:

1. 1Days 2 to 3: confirm the conversion status left Inactive, and do the first search-terms scrub
2. 2Weekly: scrub search terms, and pause any keyword that has spent meaningfully past your CAC target with zero conversions
3. 3Day 14: kill ad groups with low CTR and zero conversions; shift budget toward the campaign beating your CAC ceiling
4. 4At roughly 30 recorded purchases: switch to target-CPA bidding at your CAC target. Before that, smart bidding has nothing to learn from and burns money guessing
5. 5Scale in steps of 30% or less, only while CAC stays under your ceiling. Over the kill line: cut the worst half of keywords instead of raising budget