SaaS Metrics and Analytics

Most arguments about SaaS metrics are definition arguments wearing a costume. Two analysts look at the same books and report CAC payback of 14 months and 26 months, and neither of them is lying: one divided by gross profit and one by revenue, one counted the SDR team and one filed it under sales operations. Everything downstream inherits that fork. This page fixes the definitions first, attaches a source and a sample description to every benchmark, and then points at the page that repairs each number when it comes back ugly.

The five metric families and which decision each one settles

A metric earns dashboard space only if a specific person changes a specific behaviour when it moves. Group them by the decision they settle and the list shortens fast.

Marketing touches all five, but it only controls two of them outright. You set acquisition cost and you set pipeline creation. Retention and expansion move on product, onboarding and customer success decisions that marketing influences through content, lifecycle email and in-product messaging rather than owns. Efficiency is a composite of the other four, which is why arguing about the efficiency number without opening the four inputs is a waste of a meeting.

The formulas, and the definitional traps that make two companies incomparable

Every formula below is standard. The third column is the part nobody writes down, and it is the reason your number and a peer’s number cannot be compared until you both answer it.

Three of these traps do most of the damage. The gross margin step in CAC payback is the first. The cost inclusions in CAC are the second, and they are where internal politics live, because a marketing leader who excludes SDR cost reports a payback several months better than the same business measured properly. The third is the cohort window on retention, where including customers who signed three months ago flatters net revenue retention because they have not had time to churn yet.

Write your definitions in a shared doc, date them, and version them. When the definition changes, restate the prior twelve months on the new basis in the same chart. A metric that silently changes definition mid-year destroys more trust than a bad number honestly reported.

What good looks like by ACV band, funding stage and motion

Benchmarks are only useful inside a segment. A self-serve product at 900 dollars ACV and an enterprise platform at 120K ACV have almost nothing in common at the metric level, and averaging them produces a figure that describes neither.

Segment	CAC payback, typical	Net revenue retention, typical	Gross margin, typical	What drives the difference
Self-serve, under 5K ACV	6 to 14 months	85% to 100%	75% to 85%	Low friction acquisition, but weak expansion mechanics and high logo churn
SMB sales-led, 5K to 25K ACV	12 to 20 months	95% to 105%	72% to 80%	Human touch in the sale raises CAC before seat expansion arrives
Mid-market, 25K to 75K ACV	16 to 26 months	105% to 115%	72% to 80%	Longer cycles and committee buying, offset by real expansion paths
Enterprise, above 75K ACV	20 to 36 months	110% to 125%	70% to 78%	Heavy pre-sales cost, but multi-year expansion and low logo churn
Usage-based pricing, any band	Wide, 10 to 30 months	110% to 130%	65% to 78%	Revenue follows customer consumption, so retention compounds and margins compress

The sources behind those ranges disagree with each other, and the reason is sampling rather than error. SaaS Capital runs an annual survey of roughly a thousand private B2B SaaS companies, most of them bootstrapped or lightly funded, and its net revenue retention median sits close to 100 percent. Benchmarkit surveys a smaller panel weighted toward venture backed companies with dedicated finance functions, and reports higher figures in the enterprise cut. ChartMogul draws on product data from thousands of subscription businesses that skew self-serve and low ACV, so its retention distribution reads lower and its payback reads faster. Bessemer’s cloud work looks mostly at public companies and late stage private ones, which is the healthiest sample of all and therefore the least representative of a Series A business.

None of those is wrong. They are describing different populations. Our own SaaS marketing benchmarks set is published with the sample size, the ACV split and the definition for each field for exactly this reason, and the CAC payback benchmarks page breaks payback out by band rather than reporting a single median that flatters nobody.

Why CAC payback and net revenue retention outrank LTV to CAC

Pick two numbers to run the business on and these are the two. Payback answers a cash question with a short feedback loop. Net revenue retention answers a compounding question that determines whether growth needs to be bought every year or partly arrives on its own.

LTV to CAC is the ratio everyone quotes and the one we trust least below roughly three years of operating history. Lifetime value depends on a churn rate, and a churn rate computed on eighteen months of data on a company that has changed its ICP twice is a number with false precision. Shift monthly churn from 1.5 percent to 2.0 percent, which is well inside normal measurement noise, and LTV drops by a quarter. The ratio then moves from a comfortable 3.4 to a worrying 2.5 without anything in the business changing.

CAC payback has no such fragility. It uses money you actually spent and gross profit you actually collect. The comparison between LTV:CAC and CAC payback walks through where each one is the right instrument, and the short version is this: use payback for operating decisions inside the year, use LTV to CAC for capital allocation decisions across years once you have the retention history to support it. If you want the ratio done properly with sensitivity bands rather than a single figure, the LTV to CAC guide sets out the method.

Here is the practical difference. A team with a 22 month payback and 118 percent net revenue retention is in better shape than a team with a 12 month payback and 92 percent retention, even though the second team looks more efficient on the acquisition line. The first team keeps and grows what it wins. The second refills a leaking bucket forever, and its efficiency advantage evaporates the moment it stops spending.

Two operating rules follow. First, if payback exceeds roughly 24 months in a sales-led motion, your growth is funded by capital rather than by the business, and that is a board conversation rather than a marketing one. The CAC payback period guide covers how to calculate it without flattering yourself, and the B2B SaaS CAC payback calculator does the arithmetic including the gross margin step that most spreadsheets skip. Second, if net revenue retention sits below 100 percent, fixing it beats any acquisition improvement available to you, because every point of retention compounds annually and every point of CAC improvement does not.

How to attribute revenue to marketing without lying about it

No attribution model is true. Each one answers a narrow question and hides something specific, and the honest approach is to run two or three models that fail in different directions and report where they disagree.

Model	What it answers well	What it hides	Cost to run	Use it when
First touch	Which channels create awareness	Everything that closes the deal	Low	You are building a new demand engine and need to know what starts journeys
Last touch	Which asset preceded the conversion	All upstream demand creation	Low	You are optimising a landing page or a single campaign
Linear or time decay	A fairer spread across a long cycle	Channels with no trackable touch, like podcasts and community	Medium	Sales cycles run longer than 60 days and you have clean CRM data
W or U shaped	The three moments that matter most in B2B	Anything between the milestones	Medium	You have defined lead, opportunity and close stages properly
Data-driven in GA4	Relative weight across digital touchpoints	Anything offline or dark, and the model itself is opaque	Low	Digital-only mix with high volume
Self-reported, the how did you hear about us field	Brand, word of mouth, community, podcasts	Precision, and it decays if the field is optional	Low	Always. Run it alongside platform data, not instead of it
Geo holdout or incrementality test	Whether a channel actually caused the lift	Small channels, because the test needs volume	High	You are about to make a spend decision above roughly 500K a year

The most useful cheap instrument in the list is the self-reported field on the demo or trial form. Make it required, offer six to eight options plus a free text box, and read it monthly. It will disagree with your platform data, often by a lot: podcasts, Slack communities, peer recommendation and a colleague’s old job all show up here and in no report your ad platform will ever produce. Platform data, meanwhile, overweights the last paid click because that is the only thing it can see.

Where they conflict, the truthful statement to a board is that both are estimates and here is the range. A marketing team that says paid search contributed somewhere between 18 and 31 percent of new pipeline this quarter, depending on the model, and explains why, is more credible than one that says 24.6 percent with a straight face.

The reporting stack, and the four joins that break

The hard part of SaaS measurement is not the maths. It is stitching an anonymous visitor to a known lead, that lead to an account, that account to an opportunity, and that opportunity to closed revenue. Four joins, and each one breaks in a characteristic way.

Two honest costs. This chain takes a competent analytics engineer six to ten weeks to build for the first time, and the tooling for a mid-market company runs somewhere in the low thousands of dollars a month once you add a customer data platform, a warehouse and a BI layer. That is real money for a Series A company, and the sensible early version is a spreadsheet, a required self-reported source field and clean CRM hygiene. Build the warehouse when the spend you are trying to optimise is larger than the cost of measuring it.

The diagnostic index: a bad number and the page that fixes it

Here is the part most metrics articles skip. A number outside its band is a symptom, and symptoms have a small set of common causes. Start with the cause list before you start with tactics.

The pattern worth internalising: almost every bad efficiency number decomposes into a volume problem, a conversion problem or a price problem. Find which of the three it is before you choose a tactic. Teams that skip this step spend three months fixing conversion when the real issue was that average contract value fell 22 percent after a pricing change nobody told marketing about.

What to report to a board that does not believe attribution

Most boards have been burned by a marketing dashboard, so lead with the numbers that do not depend on a model. Pipeline created, cost per opportunity and CAC payback are all computable from spend and CRM records without any attribution assumption beyond which deals existed.

Then present contribution as a range with the method named. Something like: platform attribution puts content at 22 percent of new pipeline, self-reported survey puts it at 34 percent, and the single geo holdout we ran in the second quarter suggested the true figure sits in between. That framing survives questioning. A single precise percentage does not.

Once a year, run one real test. A geo holdout on paid search, a four week pause on a podcast sponsorship, a matched-market test for brand spend. It costs you some pipeline and it is the only evidence you will ever have that is not correlational. Budget for the loss and do it in your slowest quarter.

The quarter I stopped defending one attribution number and started showing three models side by side with the gaps explained was the quarter the CFO stopped fighting me on budget.

Reporting cadence: what you look at weekly, monthly and quarterly

Frequency should match how fast the number can actually move. Looking at CAC payback weekly produces noise and anxiety. Looking at pipeline created monthly means you find out about a problem a month late.

One more discipline: restate history whenever a definition changes. If you move from revenue-based to gross-profit-based payback in March, redraw the prior twelve months on the new basis in the same chart and say so in the footnote. The alternative is a chart with a mysterious step change that costs you an hour of the board meeting.

Start here this week

The fastest path to a trustworthy metric set is not a new tool. It is a definitions document, a required self-reported source field and one honest recalculation of payback with gross margin included.

If one number is already outside its band, start with the diagnostic index above rather than with a tactic. And if you only fix one thing this quarter, fix the definitions, because every benchmark comparison, every board slide and every budget argument you have for the next two years runs through them.