SaaS performance marketing

The best performance marketer I have worked with spent her first six weeks at a $14M ARR company writing zero ads. She spent them getting opportunity creation and closed-won revenue flowing back into Google and LinkedIn, fixing a CRM field that had been overwritten by a form-mapping rule since 2023, and getting sales to agree on one definition of an accepted lead. Cost per opportunity fell 34 percent in the following quarter. The creative never changed.

That is the job. Media buying is the visible part and the plumbing is the part that decides the outcome, which is why this page spends more time on data flow and operating cadence than on ad formats.

What separates performance marketing from demand gen and brand

Accountability to a revenue number, on a feedback loop short enough to act on. Demand generation owns pipeline creation across content, events, email and paid. Brand owns being remembered. Performance marketing owns a cost per revenue outcome that somebody can be held to in a quarterly review.

The practical test is what happens when the number goes wrong. If cost per opportunity rises 40 percent and the answer is a plan with campaign changes and a date, that is performance marketing. If the answer is a slide about brand awareness lift, it is something else, and something else may well be the right investment. It just should not be on the same scorecard.

Confusing these is how marketing teams end up arguing in circles. A CMO who asks the paid team to build brand and then judges them on cost per lead has set up a fight nobody can win, and the version of that fight I see most often involves LinkedIn thought leadership ads being killed in week five for failing a metric they were never built to serve.

The metric tree from impression to net revenue retention

Nine steps, each with a conversion rate, and you only need to watch three of them closely. The tree matters because a cost per opportunity problem always resolves to a specific step, and guessing which one wastes weeks.

The two steps that break most often are lead to sales-accepted and opportunity to closed-won. A collapsing SAL rate means the ad account is finding the wrong people, which is almost always an optimisation-target problem rather than a targeting problem. A collapsing win rate on paid-sourced deals usually means the offer promised something the product does not do, and no bid adjustment fixes that.

Below $25,000 ACV, substitute cost per activated signup and watch signup-to-paid by source instead. The band-by-band version of this sits in the SaaS advertising strategy playbook, and it matters because a metric tree built for enterprise applied to a self-serve product produces a lot of careful measurement of the wrong thing.

The eight numbers on a weekly dashboard

One screen, eight numbers, each with last week and the trailing four-week average next to it. If it does not fit on one screen it will not get looked at, and a dashboard nobody opens is worse than no dashboard because it creates the impression of oversight.

Two of those need defending. Raw counts alongside the SAL rate stop you celebrating a 70 percent acceptance rate on seven leads. And the wasted search terms list is on the weekly rather than the monthly because search term waste compounds: a broad match term burning $180 a week costs you $9,000 a year if nobody reads the report until Q4.

Everything else goes monthly. Click-through rate, cost per click, impression share, quality score, landing page conversion by device: all useful, none weekly. Watching CPC weekly produces fiddling, and fiddling resets the learning phase.

The weekly cadence: what you actually do on Monday

Ninety minutes, same slot, same order. Search terms first, creative second, pacing third, and no budget reallocations unless something is genuinely on fire.

The fourth step is the one teams skip and it is the one that pays. Reading raw opportunity records catches problems dashboards hide by aggregation, and it takes ten minutes. A media buyer who has never read a paid-sourced opportunity in the CRM is flying on instruments alone.

The monthly cadence: reallocate on cost per opportunity

Once a month, rank every campaign by cost per opportunity, move money from the bottom quartile to the top quartile, and change nothing else for the following four weeks. That constraint is the point, because the main enemy of paid performance at this stage is the team’s own restlessness.

Three rules keep the reallocation honest. Do not move more than 30 percent of any campaign’s budget in one step, because a bigger jump throws the campaign back into learning. Do not judge a campaign with fewer than ten opportunities, since the confidence interval is wider than the decision. And require a full sales cycle of data before a campaign can be killed rather than trimmed.

Month-end is also when the wider metric review happens. Pair the paid numbers with organic, lifecycle and product so that the picture is complete, which is what the monthly marketing metrics review is built to run through. Then re-forecast next month’s spend with the SaaS PPC budget calculator rather than repeating last month’s number by default.

The quarterly cadence: incrementality and kill decisions

Two jobs, both uncomfortable. Find out how much of your reported conversion volume would have happened anyway, and shut something off.

Incrementality testing means holding a channel out of a matched set of geographies or target accounts for four to six weeks and comparing total pipeline between test and control. Brand search is the obvious candidate and the results are usually humbling: a large share of brand clicks would have arrived organically, though rarely all of them, and the answer varies enormously by how crowded your competitor bidding is. Run the test with at least 40 opportunities per group or accept that you are reading noise.

The kill decision is separate and needs written criteria set at launch. A channel comes off when cost per opportunity doubles from its stable baseline across two consecutive months, when the sales-accepted rate stays under 20 percent for a quarter, or when you cannot add 30 percent more budget without cost per opportunity breaking. That last one is an inventory ceiling and it catches out Reddit, niche newsletters and small review categories regularly.

Quarterly is also when you compare against something external. Do it with your ACV band attached, using SaaS PPC benchmarks, because a $4,100 cost per opportunity is a triumph at $60,000 ACV and a slow-motion disaster at $9,000.

Who owns performance marketing at $1M, $10M and $50M ARR

The headcount question has a clearer answer than most people expect, and the surprise is that the operations side grows faster than the buying side.

At $1M ARR, resist hiring a specialist. One person running two channels well beats two people running five channels badly, and the biggest available win is usually landing pages rather than media buying. At $10M, the marketing ops hire is not optional and it is the hire most companies defer by a year. Ask any team at that stage where their conversion data lives and you will get three different answers involving Segment, HubSpot and a Google Sheet.

At $50M, the ops and analytics function is usually larger than the buying function, which is the shape of a mature program. If you are outsourcing execution instead, the selection criteria and realistic retainer bands are covered in the guide to SaaS PPC agencies. The honest version: an agency can run your channels, but it cannot own your data model, and handing them a broken one produces expensive tidiness.

The failure modes, starting with the one everybody has

Optimising to the metric you can see weekly rather than the one you are paid for quarterly. It is not a discipline problem. It is a latency problem, and it happens because the visible metric updates every day and the real one updates every quarter.

The mechanism is simple. Your bidding strategy targets form fills because form fills arrive in hours. Cost per form fill improves. Cost per opportunity quietly worsens because the algorithm found a cheaper, worse audience. Nobody notices for a quarter because the weekly dashboard says the trend is good. The fix is to get the downstream event into the platform even when volume is thin, using offline conversion import with a value assigned to each stage. Three opportunities a week is enough to start; you can weight the signal with a higher conversion value rather than waiting for volume.

A second failure is treating platform-reported conversions as truth. Google and LinkedIn both count generously and both count the same deal. Sum your platform conversion columns and compare to CRM-created opportunities in the same window: a 1.6 to 2.5 times inflation is normal and anything above three means a tracking duplicate somewhere.

Third, and quietest, is running paid without a written lead definition, which produces a permanent low-grade argument with sales and makes every number contestable. This and a dozen adjacent problems are catalogued in SaaS PPC mistakes that waste budget, and the pattern across them is the same: bad inputs, confidently optimised.

How to report paid results to a board that does not believe attribution

Report three numbers side by side and let the disagreement between them be visible rather than hidden. Pick one attribution model, one self-reported source figure, and one incrementality result, and put them on the same slide every quarter.

Platform-reported conversions go first, labelled as what they are: the ad networks’ own claim, inflated, useful only for comparing campaigns against each other. CRM-sourced pipeline goes second, using a single first-touch or last-touch rule you never change mid-year, because changing the model is how marketing teams lose credibility permanently. Self-reported source goes third, taken from a mandatory free-text field on the demo form and coded monthly.

That third number is the one boards trust and marketers avoid. Self-reported attribution consistently credits podcasts, communities, word of mouth and Google search far more than any tracking model does, and it consistently under-credits retargeting and display. Neither is wrong. They measure different things: the tracker measures the last click, the human reports what they remember. When the two disagree by more than a factor of two on a given channel, that channel is your next incrementality test.

Two practical rules make this survivable. Never report a number you cannot reproduce in front of someone, and never change the model and the budget in the same quarter, because you will not be able to tell which one caused the movement.

Why this is a plumbing job, not a creative job

Because the algorithms do the optimisation now, and they optimise toward whatever you tell them matters. Creative still matters, ad copy still matters, and neither can rescue an account pointed at the wrong event.

Here is the uncomfortable version. Two teams with identical budgets, identical targeting and wildly different creative quality will land within 15 percent of each other on cost per opportunity if both have clean downstream conversion import. Two teams with identical creative and different plumbing will land 40 to 60 percent apart. That ratio is why the first three months in any new performance marketing role should go to data flow: conversion events, CRM field hygiene, lead definition, offline import, and one agreed source of truth for pipeline.

The counter-argument deserves a hearing. In categories where the product genuinely looks like the competition, creative and offer differentiation carry more weight than the plumbing, and consumer-adjacent SaaS with a $20 monthly price lives closer to that world. If your product is a commodity to the buyer, message testing earns its place at the top of the list. For most B2B software with a sales cycle over 30 days, it does not, and the broader growth picture, including where paid sits alongside organic and lifecycle, is mapped out in SaaS growth marketing.

What to do in the next two weeks

Open your ad platforms and write down which conversion action each bidding strategy is optimising toward. Then open your CRM and count opportunities created from paid in the last 90 days. If those two things are not connected by an offline import or a conversion value, that is your entire project for the month and it is worth more than any campaign you could build.

After that, build the eight-number dashboard, book the Monday ninety minutes in the calendar as a recurring slot, and set the kill criteria for every live channel in writing before you need them. The platform-level differences that shape what each channel can actually report are covered in SaaS advertising platforms compared, and the rest of the paid operating detail sits in the SaaS PPC and paid ads hub. Do the plumbing first. The creative gets easier once the numbers mean something.