How do you measure customer-evidence volume?

Count conversations, by group, weekly. The target in a serious engagement is 60–80 customer and prospect interviews across current customers, churned accounts, closed-lost prospects and the segment you do not yet serve, and the split matters as much as the total: eighty conversations with happy customers tells you almost nothing about why anyone leaves. Instrument it with a shared document listing date, account, group and the pattern the conversation supported or contradicted. The reason this is the first indicator is that it is the only one that is entirely within the consultant's control in month one, which makes it the cleanest early read on whether they actually do the work they described in the pitch.

How do you measure roadmap decision latency?

Time from a prioritization question being raised to it being answered and communicated. Measure it in days, on a handful of real decisions, before the engagement starts so you have a baseline. In founder-led companies at $2M–$15M ARR this number is usually dominated by one queue, which is the founder's calendar, and that is precisely what a product leader is being brought in to shorten. The instrumentation is low-tech: a list of open product questions with the date each was raised and the date each was resolved. Watch for the failure mode where latency drops because decisions are being made without evidence, which is why this indicator is never read on its own.

How do you measure releases tied to a measured outcome?

As a percentage: of everything shipped this month, how much had a named metric and a stated expectation attached before it was built. Most companies at this stage start somewhere near zero and find the number uncomfortable. It is the best single proxy for whether a prioritization system is real, because attaching a metric forces someone to state what the work is supposed to do, which in turn makes it possible to decline work that cannot answer the question. The target is not 100 percent; keeping a slice for maintenance and bets is healthy. The target is that the number is known and rising.

How do you measure founder hours freed per week?

Ask the founder to track product-related hours for one week before the engagement and one week at day 75. This is the softest of the four and the one founders care most about, so it is worth the small amount of tedium. The signal is not the raw number of hours but where they moved: hours spent adjudicating priority conflicts should fall sharply, while hours spent in customer conversations may legitimately stay flat or rise. If total product hours have not moved at all by day 75, the consultant is adding a reporting layer rather than absorbing decisions.

What should the 90-day review meeting cover?

Four things, in order, and it should take an hour. First, the four leading indicators against their baselines, with the raw data available rather than summarized. Second, the written diagnosis, re-read out loud, with an honest statement of which parts turned out to be wrong. Third, the decisions the team can name that changed what they build, which is the single most load-bearing item on the list. Fourth, the lagging metrics that were nominated at the start, reviewed not for movement but to confirm they are instrumented and that everyone still agrees they are the right ones. A review meeting that spends its hour on activity rather than these four is the meeting the engagement was supposed to replace.

If total product hours have not moved by day 75, the consultant is adding a reporting layer rather than absorbing decisions.

Which metrics should you not expect to move yet?

Revenue and churn. Product decisions pay off on a lag: a segment call made in month one shows up in win rates two quarters later, and a retention fix shipped in month three cannot appear in a renewal cohort that has not come up for renewal. Expect movement in quarters two and three. Anyone promising lagging-metric movement inside 90 days is either scoping something trivial or selling, and the promise is itself a useful signal about the rest of what they told you. Name the lagging metrics at the start anyway, because the leading indicators only mean something if everyone agrees what they are supposed to predict. The full framework, including kill criteria, is in the guide to measuring fractional CPO ROI.

When should you fire them?

At day 45, if there is no written diagnosis with evidence behind it, or no interview cadence actually running. At day 60, if nobody on the team can name a decision that changed what they build. At any point, if the recommendations require buying more of their time to understand, or if you find yourself managing them rather than the reverse. Put these in the agreement before day one. Sivan Kadosh, a fractional Chief Product Officer for B2B SaaS companies between $2M and $15M ARR, works on a $8,000 monthly retainer with a 3 month minimum, and the stop criteria are written into the scope rather than negotiated later. Product management consulting for B2B SaaS is the category; the fractional CPO model is the version of it that accepts these measures, because it is the version accountable for the outcome rather than the analysis.

Four leading indicators carry the first 90 days: evidence volume at 60–80 interviews, decision latency, releases tied to a measured outcome, and founder hours freed. Revenue and churn are named at the start and reviewed in quarters two and three. Agree all six before the start date, along with the day 45 stop criteria, and the review meeting writes itself.