The Metrics Spine: Volume, Conversion, Time

The first quarter I owned marketing operations, I spent a week trying to reconcile two funnel reports that could not both be true. Marketing’s dashboard showed roughly 80% of sales meetings turning into opportunities. The sales team swore most of their meetings went nowhere. Both were pulling from the same CRM. When I finally traced one quarter’s MQL cohort forward, account by account, the real rate was closer to 30%, and only about a third of the “meetings” had any event logged behind them at all. Neither team was lying. They were measuring different things and calling them the same word, and no dashboard reconciles two definitions that were never agreed on in the first place.

That is the problem this chapter solves. The last chapter gave us the Bowtie: eight sub-systems, one lifecycle, Mutual Commit at the knot. A shape is not an instrument. To run the factory you have to measure every stage of it the same way, in a language that holds across marketing, sales, and customer success, and across every motion you run. Revenue Architecture gives that language three words: volume, conversion, and time. Everything you will ever put on an operating dashboard is one of those three, stamped with the Bowtie stage it belongs to.

Three families, one grammar

Volume metrics, written VM, count how many of something you have at a point in the journey: prospects who fit the profile, MQLs, wins, committed MRR, lifetime value. Conversion metrics, written CR, measure how efficiently one stage feeds the next, and they are always a ratio: CR = output / input. Time metrics, written Δt, measure the wait between one conversion and the next, not how long an activity takes but how long a deal or a customer sits before it moves. A win rate of 22% (CR) is a different fact from 40 committed deals (VM), which is a different fact from a 47-day sales cycle (Δt). You need all three to describe a single stage honestly, and most arguments about “why the number moved” are really two people holding two different families.

The taxonomy runs the full length of the Bowtie. Number the stages 1 through 9 and each gets its own VM, CR, and Δt. The point of numbering rather than naming is comparability: VM4 means “verified priority, inaction now carries a consequence” whether you call it a SAL in an SMB motion, a qualified opportunity in enterprise, or a product-qualified account in a self-serve motion. Same slot, different label, one definition.

Two boundaries in that table earn a note. The left side (VM1 to VM5) is counted in units, people and deals; the right side (VM6 to VM9) is counted in money, MRR and ARR and LTV. The hinge between them is VM6 = VM5 x ACV, the moment the factory stops counting logos and starts counting revenue. And CR9 stays reserved for the closed loop, the feedback path the course builds out later. Leave the slot empty; do not invent a metric to fill it.

Conversion rates compose

The reason to number conversions instead of naming them is that they multiply into the ratios executives actually ask for. Four identities do most of the work:

L2O = CR1 x CR2        lead to opportunity
OTC = CR3 x CR4        opportunity to close
NRR = CR7 x CR8        retention x expansion
GRR = CR7              retention alone, capped at 100%

Once the stage rates are clean, the composites fall out of them, and a bad composite tells you which stage to open up. If lead-to-opportunity sags, you look at CR1 (are we generating the right leads?) before CR2 (are we working them well?). GRR and NRR both leaning on CR7 is the tell that expansion can only ever add to a retention base you have already kept. You cannot expand your way out of churn for long, because CR8 multiplies a number that CR7 has already shrunk.

Composing also shows you which stages you can trust. In the funnel I instrument now, the opportunity-to-qualified step (SAO to SQO, both stamped on the opportunity record) holds near 75% quarter after quarter, which makes it a stable rate to plan against. The MQL-to-meeting step swings between 8% and 21% depending on the quarter, partly because meetings are captured badly upstream. A composite built on a shaky stage inherits the shake.

Time is a lever, not a footnote

Revenue Architecture splits the Bowtie into two commands: Go Fast on the left, Go Long on the right. On the acquisition side, shorter is better, because acquisition runs on frequency and frequency is one over time. Compress Δt4, the sales cycle, and you get more cycles per rep per year out of the same headcount, no new hires required. On the retention side, longer is the whole game, because value compounds with duration. Stretch Δt8, the customer lifetime, and LTV climbs without a single new logo.

Two time metrics move the system out of all proportion to their size. Δt1, inbound response time, decides how many leads ever convert at all: the Lead Response Management Study found a lead contacted within five minutes is about 21 times more likely to qualify than one contacted at thirty. That is not a coaching tweak, it is an order of magnitude, and it is why the first operations reflex on any inbound motion is to attack response latency before touching anything else. Δt6, time to first impact, is the mirror image on the right side. The longer a new customer waits to feel the outcome you promised, the more likely they churn before their first renewal. Δt1 governs whether the deal starts; Δt6 governs whether the revenue survives. Instrument both before you polish the metrics in between them.

Ratios that cross domains

Some numbers do not belong to a single stage. Performance metrics are ratios that transcend the Bowtie: revenue per rep, average price per deal (committed MRR divided by deals), win rate held up against retention. That last pairing is a quiet diagnostic. A rep with a high win rate and mediocre retention is often overselling, closing fits that do not last. Revenue Architecture makes the point with two sellers working the same fifteen opportunities a month. Rob wins 33% but retains 90% (0.9 NRR). Jennifer wins 20% but retains 100% and expands to 1.2 NRR. Over five years Jennifer’s book is worth about $1.47M against Rob’s $1.11M. Higher NRR beats higher win rate over a lifetime, and a spine that measures only the left side of the Bowtie will crown the wrong rep and comp the wrong behavior.

Leading and lagging

Sort every metric by which side of the Bowtie it sits on and you have sorted leading from lagging. Left-side volumes and conversions (leads, meetings, CR1 through CR4, the response times) lead: they move first, and you can act on them this week. Right-side outcomes (GRR, NRR, LTV, and the delayed NPS survey) lag: they confirm months later what the leading signals already predicted. You steer on the leading metrics and you are graded on the lagging ones. Running the business off the lagging numbers is driving by the rear-view mirror, because by the time GRR drops, the customers who left did so two quarters ago.

Benchmarks, and the better benchmark

A standardized spine unlocks comparison, and there are three reference standards worth naming. You can benchmark against your industry (Revenue Architecture publishes a SaaS set spanning CR1 through CR8 by ACV band, drawn from 868 companies), against named peers, or against yourself over time. External benchmarks are useful as a sanity check and dangerous as a target. A median net revenue retention hovering near 101% in recent B2B studies tells you where the pack sits, not what your motion should hit.

Self-benchmarking is the one that changes decisions. Track your own rates as a quarterly trendline and small moves show their real cost.

The spine is a contract, so govern it

None of this survives a growing organization unless the definitions hold still. The fastest way to break a metrics spine is to let every team quietly redefine a stage. One group counts an SAO before qualification, another after. Someone reports on a field that floors at zero and loses every downgrade. A well-meaning admin adds a second ARR field next to the real one, and half the reports drift onto the wrong column. Revenue Architecture is blunt about the fix: once the data model is agreed, changes to it run through a review, because a change to a metric definition is a change to how the business is valued.

That governance is a short, boring, load-bearing process.

Ground it in whatever your reporting field guide already settles. In mine, new-business pipeline means exactly New_ARR__c > 0, SAO precedes SQO by definition, and a short list of tempting fields is banned because they lie on downgrades. Those are not preferences, they are the contract that lets a marketing number and a sales number roll up to the same total. For the deeper stewardship playbook, the data governance topic walks it stage by stage; for wiring these definitions into the CRM objects themselves, the engineering academy’s data-modeling track picks it up.

Handing off

You now have the spine: volume, conversion, and time by Bowtie stage, composing into the ratios leadership asks for, sorted into signals you steer by and outcomes you are judged on, and pinned down by a change-request process so the definitions survive the next reorg. What you do not have yet is the math that runs on top of it. Why does a five-point win-rate gain sometimes double revenue? Why does one more point of NRR bend the LTV curve so hard? Those are not conversion questions, they are questions about how these metrics multiply and compound. That is the Mathematical Model, and it is next.