A CEO showed me his company dashboard last month. Forty-one tiles. Signups, trials, MQLs, SQLs, activities logged, emails sent, demo requests, page views, NPS, pipeline created, pipeline coverage, ARR, MRR, ACV, logo churn, revenue churn, and a tile called "Momentum" that nobody could explain to me. Nine million ARR, 55 people, Series B closed eighteen months ago.
I asked him one question. What is your net revenue retention?
He pulled it up. 118%. Then his VP of Finance said it was 104%. Then someone from customer success said the number they report in QBRs is 111%. Three people, three answers, same company, same quarter. All three numbers were computed correctly. They just answered three different questions, because nobody had ever written down which question they were asking.
That is the actual state of SaaS metrics at most companies between $2M and $30M ARR. The dashboards are full and the numbers are meaningless. And every time I run a RevOps audit, the fix is the same shape: delete most of the tiles, define the ones that survive, and rebuild them from fields the CRM can actually produce.
The formula was never the problem
Search "SaaS metrics" and you get a hundred articles that hand you formulas. NRR equals starting ARR plus expansion minus contraction minus churn, divided by starting ARR. Fine. Nobody I have worked with was confused about that.
What they were confused about was everything underneath it. Does a customer who downgraded in March and upgraded in August count as expansion, contraction, or both? Does a one-time onboarding fee sit inside ARR? What about the three customers on a legacy annual contract that auto-renews at a discount you no longer offer? What happens to a logo that churned from one entity and resigned under a different one after an acquisition?
None of that is math. All of it is data modelling, and it lives in your CRM and billing system, not in a spreadsheet formula. A metric is only as good as the field that feeds it, which is why teams with bad CRM data quality cannot be rescued by better dashboards. You are averaging garbage more attractively.
A metric your team defines three ways is worse than no metric at all.
No metric produces honest uncertainty. Three definitions produce confident, contradictory decisions, and the loudest number usually wins.
Track eight numbers
Here is my actual opinion, and it will annoy some people. A B2B SaaS company under $30M ARR needs eight metrics on the board and executive view. Not fifteen. Eight. Everything else is a diagnostic you pull when one of the eight moves.
1. ARR growth rate, year over year. The headline. Not month over month, which is noise at your size, and not a rolling quarterly annualised figure, which flatters you in a good quarter and terrifies the board in a bad one.
2. Net revenue retention. Expansion minus contraction minus churn, on the existing customer base only. This is the one number that tells you whether your business compounds without sales effort. SaaS Capital's 2026 survey of more than 1,000 private B2B SaaS companies found that moving from the 90 to 100% NRR band into the 100 to 110% band came with roughly five percentage points of extra growth. It is the number with the most upside per point you own. More detail in our NRR guide.
3. Gross revenue retention. NRR without expansion. This is the honest one. NRR of 115% sounds great until you learn it is built on 88% GRR and a handful of aggressive upsells papering over a leaky base. Track both or track neither. Our GRR guide covers why boards increasingly ask for this first.
4. CAC payback period in months. Fully loaded sales and marketing cost divided by new ARR times gross margin. The single best test of whether growth is affordable. See CAC payback for the version that includes the costs most teams quietly leave out.
5. Gross margin. Boring, and the reason half of "SaaS" companies are actually services companies with a login page. If it is under 70% and you are calling yourself SaaS, the metric is telling you something the pitch deck is not.
6. Burn multiple. Net cash burned divided by net new ARR added. Under 1.5 is healthy at growth stage. Above 3 means the engine is consuming more than it produces and no amount of pipeline commentary changes that.
7. Win rate, segmented. One blended win rate is useless. Win rate by segment, by source, and by whether an incumbent was present tells you where to put next quarter's headcount. Our guide on calculating and benchmarking win rate covers the segmentation that makes it actionable.
8. Qualified pipeline coverage for the next two quarters. Not this quarter. This quarter is already decided. The number that predicts your problems is coverage on the quarter after next, and most teams never look at it until it is too late to fix. We wrote about why the 3x coverage rule quietly lies.
That is the list. Four tell you about the business model, four tell you about the go-to-market engine. Everything else you might want is a drill-down from one of these.
What to actually track at your stage
The eight above are the ceiling, not the starting point. Trying to run all eight at $800K ARR is theatre. You do not have enough customers for NRR to mean anything, and your CAC payback is computed on a sample size of nineteen.
The mistake I see most often is a Series A company adopting the metric pack of a Series C company because a board member shared a template. You end up with a beautiful reporting cadence built on samples too small to be real, and the team learns to distrust the whole exercise. That distrust is expensive and takes about two years to undo.
The tiles I would delete from your dashboard
Some of these will hurt. I stand by all of them.
MQL count. A number your marketing team controls the definition of, reports on, and is compensated against. Track MQL to SQL conversion rate if you must track anything in this family, because a ratio is harder to inflate than a count. Better still, track qualified pipeline in dollars.
Activities logged. Calls made, emails sent, tasks completed. This measures compliance with your CRM, not selling. The day you put it on a dashboard is the day it stops being informative, because reps optimise for what is measured and logging an email costs nothing.
Total leads and signups. Volume without qualification. A number that goes up when you buy a bad list.
NPS. I have written about why B2B NPS is mostly noise. Response rates under 20%, self-selected respondents, and a scoring method that treats a 6 and a 0 identically. It is a customer success ritual, not a metric.
LTV as a standalone figure. LTV requires you to assume a churn rate that continues forever. At Series A you have maybe eleven quarters of data. The number is a projection wearing a fact costume. Use it inside LTV to CAC if your board asks, and be honest that the denominator is the only half you know.
Rule of 40, under $10M ARR. Genuinely useful later. Below $10M it mostly tells you that you are growing fast and burning cash, which you already knew. Our Rule of 40 guide covers when it starts earning its place.
Anything called "score" that you built in a spreadsheet and cannot explain in one sentence. Health scores, momentum scores, engagement indexes. If the composite cannot be reconstructed by the person reading it, it is decoration.
Definition drift is the actual bug
Back to the CEO with three NRR numbers. Here is what had happened, and it happens everywhere.
The metric dictionary is unglamorous and it is the highest-return two days of work in RevOps. One page per metric. Name, plain-English question it answers, exact formula, source system, owner, refresh cadence, and a short list of decided edge cases. Ten metrics, ten short entries.
Write down the ugly ones specifically. Does a customer who pauses for two months and returns count as churn and new, or as continuous? Do you count ARR on a signed contract that starts next quarter? Is a multi-year deal counted at year one value or annualised across the term? There is no universally right answer to any of those. There is only a decision, applied consistently, that you do not relitigate every board cycle. Baker Tilly's diligence teams see valuation haircuts from exactly this: ARR that cannot survive investor scrutiny because the definition moved between reporting periods.
If your ARR, bookings, and recognised revenue numbers are currently interchangeable in conversation, start with our guide on ARR vs bookings vs revenue. Separating those three is usually what makes everything downstream work.
The CRM fields behind each metric
This is the part the metrics listicles skip, and it is where the work actually is. You cannot report NRR if the CRM cannot tell you what a customer paid at the start of the period. Most cannot, because the deal record holds a one-time contract value and nothing tracks the subscription state over time.
The minimum data model for the eight metrics looks like this. A subscription or contract object separate from the deal object, holding ARR value, start date, end date, and a status. A change log on that object so an upgrade in month seven is a dated event rather than an overwritten field. A close reason and a competitor field on lost deals, required at close, because optional fields on lost deals are filled in roughly 11% of the time. A segment field set at the account level and never inferred from employee count at query time. A pipeline stage definition with an entry criterion the rep cannot fudge.
In HubSpot this means custom objects for subscriptions rather than stretching the deal record until it snaps, plus a properly designed reporting layer on top. In Salesforce you have the contract and asset objects already and most teams ignore them. Either way the pattern holds: metrics are a modelling problem first and a reporting problem second.
We build this data layer as the first phase of most CRM and RevOps engagements, because building dashboards before the model is right just produces prettier disagreement. The automation work comes after, once there is something reliable to automate against.
Benchmarks for 2026, and how to use them
Numbers help, so here are current ones. Treat them as orientation, not targets.
SaaS Capital's 2026 survey puts the population median growth rate at 22%, down from 25% in 2024, with bootstrapped companies at 20% and equity-backed at 25%. Their spending benchmarks show bootstrapped companies spending 12% of ARR on sales and 6% on marketing, against 20% and 12% for equity-backed peers. Total median spend runs at 96% of ARR for bootstrapped and 101% for equity-backed, which is a polite way of saying the median venture-funded SaaS company still spends slightly more than it makes.
NRR varies enormously by who you sell to. Enterprise at over $100K ACV sits near 118% median, mid-market around 108%, and SMB under $25K ACV around 97%. If you sell to SMB and your board is asking why you are not at 120% NRR, that is a segment conversation, not a performance conversation.
Two warnings about benchmarks. First, medians hide bimodal distributions, and SaaS is deeply bimodal right now. Second, a benchmark is only comparable if the other company defined the metric the way you did, and you already know how unlikely that is. Use benchmarks to notice when you are wildly off, not to set quarterly goals.
How to fix this in about a week
Not a quarter. A week, if you accept that version one will be imperfect.
Day one, pick the metrics. Use the stage table above. Resist adding a ninth because someone likes it.
Day two, write the dictionary. One page each. Decide the edge cases with the CFO and the head of CS in the room, then stop discussing them.
Day three and four, fix the data model. Add the subscription object, the change log, the required close reason, the account-level segment field. Backfill what you can and mark what you cannot, because a known gap is workable and a silent gap is not.
Day five, build one dashboard. One. Eight tiles, each linking to its definition. Kill the other dashboards or move them into a folder marked archive so nobody quotes them by accident. Our sales dashboard guide covers the layout that executives actually read.
Then leave it alone for a quarter. The temptation to add tiles is constant and every addition costs you a little trust in the whole thing.
Three people, three answers, one metric?
We will audit your metric definitions and the CRM fields behind them, then show you which numbers your board should stop trusting. Free 30-minute session.
Book an audit →FAQ
What are the most important SaaS metrics for a B2B company?
Growth rate, net revenue retention, gross revenue retention, CAC payback period, gross margin, burn multiple, segmented win rate, and forward pipeline coverage. Four describe the business model and four describe the go-to-market engine. Under $5M ARR you only need the first four to be meaningful, because the others are computed on samples too small to trust. Everything beyond these eight is a diagnostic you pull when one of them moves in a direction you did not expect.
How many SaaS metrics should we report to the board?
Eight on the standing slide, with a written definition attached to each. Boards ask for more because more feels like rigour, but a longer pack reliably produces shallower discussion. The better move is a short pack plus a deep dive on one metric per meeting, rotating quarterly. If a board member wants a number that is not on the list, that request is a good prompt to ask what decision the number would change.
What is a good net revenue retention rate in 2026?
It depends almost entirely on your segment. Enterprise SaaS above $100K ACV runs around 118% median, mid-market around 108%, and SMB under $25K ACV around 97%. Median across all private B2B SaaS has compressed to roughly 101 to 103%. Anything above 110% in mid-market is genuinely strong. Chasing a headline NRR number without stating your segment is how teams end up feeling bad about a healthy business.
Why do different teams in our company report different ARR?
Because they are pulling from different systems and applying different rules, and nobody wrote the rules down. Finance usually computes from billing, sales from closed-won deal amounts, and customer success from account records. Each is internally consistent and they will never agree, because deals record a moment while subscriptions record a state. Fix it by naming one system of record for ARR, deriving the number from subscription records rather than deals, and documenting the edge cases once.
Should we buy a SaaS metrics tool or build the reporting ourselves?
Under $10M ARR, build it in your CRM. The reason most metrics tools disappoint is that they read from the same broken data model you already have, so you pay a subscription for faster wrong answers. Fix the model first, then decide. Above $10M ARR with a billing system, a usage database, and a CRM that need joining, a proper reporting layer starts to earn its cost. The question is never which tool. It is whether your subscription data is modelled well enough for any tool to read.
The uncomfortable truth about SaaS metrics is that the reporting problem is almost always a data modelling problem wearing a BI costume. Every team I have worked with that fixed its dashboards by buying better dashboards was back in the same place nine months later, with a nicer-looking version of the same disagreement.
Cut to eight numbers. Write down what each one means. Build the fields that produce them. If you want a second pair of eyes on which of your current metrics would survive contact with a diligence process, get in touch and we will start with your ARR definition, because that is usually where the trouble begins.