Performance review template for B2B SaaS

The performance review with the columns a subscription business needs: every line reviewed by ICP segment and funnel stage, results read at the point the cycle allows, and a cohort date so a period is compared with one that has had the same time to mature.

Everything else is the general review, unchanged. Planned from the plan, variance separated from cause, cause held to evidence, one decision with one owner and a date.

Run it monthly on the inputs and quarterly on the outcomes, and be explicit about which one is happening. A monthly review that judges pipeline is judging an immature cohort; a quarterly review that ignores the inputs leaves eleven weeks with nothing to act on.

Run the quarterly one at least a full cycle after the spend it covers.

The single most useful sentence at the top of a review in this category is which kind of review it is.

An input review reads the things that actually move week to week: reach into the named segment, how much of the list has been covered, cost per lead, how quickly leads were handled, whether the creative is reaching the right titles. Those are all real, all actionable, and none of them is the outcome.

An outcome review reads pipeline, and it can only honestly be run on cohorts old enough to have matured. In most businesses that means a quarter behind the spend it covers.

Teams get into trouble by running one meeting and treating it as both. The pipeline numbers from the last six weeks are read as though they were the result, they always look poor because the cohort is young, and a working campaign is cut.

The cohort date is the column that makes comparisons legitimate.

A cohort six weeks old and one six months old are not comparable. The younger one has had less time for deals to be accepted, progressed or closed, and it will look worse on every measure that depends on time. Line them up in a table and the conclusion is always that performance is declining.

So a review compares cohorts at the same age: this quarter's cohort at twelve weeks against last quarter's cohort at twelve weeks. It is more work to assemble and it is the only version that supports a decision.

Where the data cannot be cut that way, say so in the notes and treat the comparison as indicative rather than as evidence.

The unit is the ICP segment crossed with the funnel stage, matching the plan.

Segment matters because a lead's value differs: a lead from a segment with a large deal size can cost several times what a small-segment lead can and still be the better purchase. A review that compares cost per lead across segments without their deal sizes will recommend defunding the most valuable one.

Stage matters because the events differ. Awareness is judged on reach into a named market and on what happens to capture volume later. Capture is judged on cost per accepted lead now. Holding them to one table is how upper-funnel spend gets cut in every business that does it.

Three causes recur and are worth naming explicitly, because each looks like something else.

List depletion. When most of a named list has been reached, frequency rises and cost per lead climbs. From inside the platform it looks exactly like creative fatigue, and new creative will not fix it.

Handover quality. Leads that were called late, or by the wrong person, or during a week nobody was warned about, convert worse for reasons that have nothing to do with the advertising. Where the handover was poor, the row should say so rather than attributing the result to the campaign.

Definition drift. A tightened acceptance definition makes every subsequent period look worse and every earlier one look better. Record the change with its date, and treat the periods either side as different measurements.

Four decisions cover most rows here.

Continue, which for an awareness row often means continue for another quarter because the evidence is not in yet, and saying that explicitly protects it.

Change the weight or the segment, where the evidence supports a comparison at equal maturity.

Change the list, where depletion is the cause. This is usually a project rather than an edit, and the row should say so.

Investigate, where the cause is not known. In this category the investigation is frequently a join between advertising and pipeline data that nobody has done, and that is a legitimate and valuable decision.

The notes column carries the events that make a period unrepresentative, and in this category they are often organisational rather than market-driven.

A sales reorganisation. A pricing change. A product launch that moved. A quarter-end push that pulled deals forward. A departure that left a segment uncovered for a month. Each of those moves the numbers by more than most campaign decisions do.

Write them down at the time. A quarter's poor result with no recorded context becomes, a year later, evidence about a segment that was never at fault.

The individual reviews change next quarter. The accumulation answers the questions a demand generation team is otherwise guessing at.

Which segments have ever produced pipeline, at what cost, and how long each one took to show it. How much of the list in each segment has been covered, and how quickly it depletes. Whether awareness spend has ever visibly moved capture volume, and on what lag. What the acceptance rate is by campaign, over enough periods to be trusted.

None of those can be answered from a single quarter, and all of them are answerable from four reviews written the same way. Keep them together, by segment, and read the previous ones before writing the next.

Seven checks. The review says at the top whether it is judging inputs or outcomes. Every row is reviewed by ICP segment and funnel stage. Outcome rows cover cohorts with a full cycle behind them, compared with cohorts of the same maturity. List depletion, handover quality and definition drift are considered as causes before the campaign is blamed. Every decision is one of four, with the change named. Every decision has an owner and a next-review date. And organisational events that moved the numbers are recorded while they are still remembered.