When you have a sales team and nobody owns the choice of what to fix next, week to week, a fractional CRO for startups can take it on.
Your next revenue improvement starts with a choice
Choosing your next revenue improvement means deciding where a limited week of work can make a useful difference.
You could spend it finding more opportunities. You could improve evaluations, help buyers get a contract approved, or investigate why recent customers have stopped using the product. Each sounds sensible. You cannot give everything your full attention.
The tempting shortcut is to open the funnel report and fix whichever percentage looks worst. At least that gives you somewhere to start.
But the smallest percentage tells you where the largest share drops away. It does not tell you whether those buyers should progress, what would help them, or how much work that help requires.
One product builder listed product need, positioning, audience and onboarding as possible problems, "or simply no effective distribution channel". 11 Product builder's account of uncertainty about product need, positioning, audience, onboarding and distribution.https://www.reddit.com/r/startups/comments/1v36bb6/70_users_tried_my_product_not_one_stuck_around/ Several explanations could fit the same disappointing result. The work was choosing which one to investigate.
I would start by making those explanations easier to compare. What happened to the people who entered? Where did their progress change? What would have to become true for more of them to buy?
The purpose of measuring a revenue engine is to make your next choice better. A larger report is only useful if it helps you make that choice.
Follow one group before comparing conversion rates
A useful conversion rate follows the same opportunities from one confirmed step to the next.
A pipeline snapshot answers a different question: what is sitting here today? Deals in evaluation may have arrived earlier than deals in discovery. Dividing those two stage counts does not reveal how many discovery opportunities progressed.
Start with a cohort, meaning one comparable group. Choose one product, buyer segment and selling motion. Then select every opportunity that entered during a fixed window, using the same entry rule throughout.
Keep unsuccessful opportunities in that original group. Removing them changes the denominator and makes the process look better without changing what happened.
For each step, count how many members of the group reached it. Divide the number reaching the next step by the number reaching the previous step. Count each opportunity once at each reached step.
Suppose your steps are nested: everyone who signs has accepted the solution, and everyone who accepts it has agreed to evaluate. Multiplying the step rates then gives the group's overall conversion rate, which is its win rate.
Each rate follows the people who reached that point. You are tracing one group all the way through, rather than multiplying unrelated percentages.
Real buying paths can be less tidy. If buyers skip a step, record the skip and decide how that route will be measured. Do not invent an evaluation to make the funnel balance.
If a buyer returns, preserve that history without counting the same opportunity twice in the original group. Write down your counting rule so someone else can reproduce the result.
Keep the entry rule stable while comparing groups. A stricter definition of an opportunity can raise conversion without improving anyone's buying experience.
Measure progress the buyer has actually confirmed
Buyer-confirmed progress makes your measurements useful for understanding what a purchase still needs.
“Demo delivered” describes something you did. “Buyer agreed an evaluation” describes a decision the buyer made. Both can belong in your records, but they answer different questions.
For this chapter's example, the path is:
Problem confirmed → Evaluation agreed → Solution accepted → Contract signed
Use steps that fit the actual purchase. Some products need a trial, some need procurement approval, and some have a much shorter path. The labels matter less than the evidence behind them.
Write down what would establish each confirmation. A friendly response is not necessarily agreement to evaluate. A request for pricing does not, by itself, establish that the buyer accepts the solution.
Stage exit criteria take that definition work further. Here, the aim is to compare possible improvements across the group.
Record the date of entry and each confirmation. Also record whether the opportunity remains open, ended without a contract, or signed.
Time changes what a count means. A recently admitted group may have fewer contracts simply because buyers have had less time to decide. Compare groups at the same age, measured from entry.
For a group admitted across several dates, use the same follow-up period for each member. An observation date should make clear who has reached that age and who still needs more time.
An open opportunity is unfinished evidence. Report it as open, rather than treating it as a loss or quietly leaving it out.
Show days from entry to each confirmation and signature. Keep the open count beside completed-deal timing, so a fast handful of wins does not stand in for the whole group.
The smallest percentage can offer the smaller gain
The most valuable improvement depends on the change you can achieve, not just the rate you start with.
Consider a hypothetical group of 100 similar new-business opportunities. Each entered when the buyer confirmed a problem. For this teaching example, every opportunity is resolved by its 180-day review.
Of the original 100, 40 agreed an evaluation, 20 accepted the solution and 10 signed. The counts are nested, and every loss remains in the original group.
The conversion rates are therefore:
- Problem to evaluation: 40 ÷ 100 = 0.40, or 40%.
- Evaluation to acceptance: 20 ÷ 40 = 0.50, or 50%.
- Acceptance to contract: 10 ÷ 20 = 0.50, or 50%.
Overall conversion is 10 ÷ 100 = 0.10, or 10%. Multiplying 40% × 50% × 50% gives the same answer.
Assume every signed contract has a fixed value of £10,000. The result is £100,000 in contract bookings. That is signed business, not recognised revenue or cash received.
Now compare two hypothetical changes for a future comparable group of 100 opportunities, observed over the same follow-up period.
| Hypothetical scenario | Evaluation agreed | Solution accepted | Contract signed | Expected contracts | Modelled bookings |
|---|---|---|---|---|---|
| Baseline | 40% | 50% | 50% | 10 | £100,000 |
| A: improve evaluation agreement | 50% | 50% | 50% | 12.5 | £125,000 |
| B: improve contract completion | 40% | 50% | 70% | 14 | £140,000 |
Each percentage is conditional on reaching the preceding step. For A, the calculation is 100 × 0.50 × 0.50 × 0.50 = 12.5 expected contracts. For B, it is 100 × 0.40 × 0.50 × 0.70 = 14.
A fractional contract belongs to the model. Actual contracts arrive as whole signed agreements.
B produces the larger modelled gain while improving a step that already converted better. That is enough to reject “always fix the lowest rate” as a decision rule.
It is not enough to choose B. Its assumed improvement is larger, and nothing in the arithmetic establishes that buyers will respond that way.
Put the work and the deadline beside the gain
A worthwhile improvement must fit the people and time available to deliver it.
Compare the two hypothetical improvements again: A helps buyers agree an evaluation; B helps accepted solutions reach signature. Suppose A needs 30 extra seller hours across the next group and B needs eight. Assume room to deliver four additional contracts during the relevant period.
| Decision input | Hypothetical A | Hypothetical B |
|---|---|---|
| Extra expected contracts | 2.5 | 4 |
| Extra modelled bookings | £25,000 | £40,000 |
| Extra seller hours | 30 | 8 |
| Expected volume fits four assumed delivery places | Yes | Yes, with no spare room in the model |
B deserves investigation first on expected bookings and selling effort. Whether you can deliver the work still needs checking.
Four expected extra contracts is not a ceiling on how many buyers might sign. For either option, check whether you could serve five extra buyers, or four arriving in the same week. If you cannot, agree feasible start dates before committing delivery time.
If B only raises the final rate to 55%, the calculation becomes 100 × 0.40 × 0.50 × 0.55 = 11 contracts. Modelled bookings fall to £110,000.
A now has the larger modelled gain. Whether it is worth the extra hours depends on your priorities and available capacity.
The timing matters just as much. These sums describe a future group over a comparable observation window. They are not a forecast for next month's bookings.
Write down when the change can reach buyers and when those buyers could reasonably sign. If you need cash sooner, also check the assumed billing and payment dates. Signed value alone cannot answer a cash question.
Check what each contract requires after signature. Discounts, poor-fit customers or extra unpaid delivery work can make a higher win rate a worse business result.
The choice should change when credible assumptions change. That is the point of putting the comparison on paper: you can see what would reverse your decision.
Buyer reasons turn a calculation into a test
Buyer evidence determines whether a modelled improvement deserves a real experiment.
Return to B, the proposed improvement between acceptance and signature. The group tells you where purchases ended. It does not tell you why.
Review the unresolved questions behind those outcomes. Was the solution accepted by someone who could move the purchase forward? Was the buying process understood? Did the buyer still have a reason to act?
Treat those as questions, not a diagnosis. Look for buyers' own explanations and the dated evidence in your records. Missing notes are a reason to investigate, not permission to assign a cause.
If the records suggest confusion about approval steps, a candidate test might be agreeing those steps earlier with relevant buyers. You would then watch whether buyers confirm the route and complete the agreed actions.
The proposed test checks whether clarifying approval steps helps buyers move forward. No increase in contracts has been demonstrated; that result takes longer to observe.
The deals stuck in pipeline chapter helps with the next question on an individual purchase. Your task here is to decide whether a recurring issue deserves attention across the group.
Keep the full group in the calculation even when a few detailed reviews supply the questions. An interesting lost deal is not a conversion-rate denominator.
With a small group, show counts beside percentages. One additional signature can change a rate sharply. Give that result room to develop before turning it into a precise target.
Finally, remeasure the entire path. A change that attracts more evaluations could alter later conversion, contract value or workload. Holding those quantities constant was a modelling assumption, not a promise about the business.
A faster sales process still needs a useful outcome
Established revenue models already connect the quantities this decision needs.
Salesforce defines sales velocity as opportunities multiplied by deal value and win rate, divided by sales-cycle length. 22 Salesforce, sales velocity formula.https://www.salesforce.com/blog/sales/sales-velocity/ That is a useful way to connect output with time.
The calculation does not, by itself, choose between improving evaluation agreement and helping accepted solutions reach signature. You still need an attainable change, a buyer explanation and an estimate of the work.
Seller capacity deserves attention too. Salesforce reports that the average seller spends 40% of their time selling, from a survey conducted during August through September 2025. 33 Salesforce, State of Sales report announcement, 2026. Survey conducted August through September 2025.https://www.salesforce.com/news/stories/state-of-sales-report-announcement-2026/
That survey does not establish how your team spends its week. Use it as a prompt to inspect workload, rather than a reason to buy automation or assume administration is your biggest problem.
There is also a good argument for fixing the lowest conversion rate. If buyer evidence supports a cheap, achievable improvement there, it may be your best choice. The method should allow that answer.
The aim is to replace an automatic choice with a reasoned one. A modest improvement you can deliver may beat a larger gain that exists only in the spreadsheet.
Signed contracts are one part of the engine
A complete revenue engine includes what happens after the customer signs.
Winning by Design's Bowtie connects acquisition, retention and expansion in one view. It also explicitly uses the lifecycle to identify bottlenecks. 44 Winning by Design, Bowtie: acquisition, retention, expansion and bottleneck identification.https://winningbydesign.com/bowtie/ Connected revenue measurement is established territory.
This introduction starts with acquisition because it gives you a manageable decision to work through. A business losing customers or struggling to deliver may need to begin after signature instead.
The chapter on sales forecasting accuracy addresses prediction against actual results. Pipeline hygiene addresses which opportunities deserve work. This exercise helps you choose an improvement across comparable opportunities.
Those are related decisions, but each needs a clear question. Here, the question is simple: what should we try next, and what evidence would make us continue?