Phase plan template for B2B SaaS

The phase plan with the limits a subscription business needs: authority bounded by funnel stage rather than by campaign, upper-funnel spend protected from short-horizon decisions, and evidence conditions written for a cycle that takes a quarter to report.

Everything else is the general phase plan, unchanged. One row per area per phase, an explicit approval list, a reversal rule and a record of every change.

Write it when demand generation is run by somebody other than the person who owns the number, which is most teams. The long cycle makes both extremes worse here than elsewhere: approvals delay changes that needed to happen weeks ago, and unbounded authority produces drift nobody can trace.

Revisit it at each quarterly review, alongside the allocation.

In a business where the outcome of this month's spend appears next quarter, the two usual arrangements fail harder than they do elsewhere.

Approval for everything is slower than the account needs, and the delay is not obvious: a change that should have been made in week two is made in week five, and its effect is not visible until the quarter after the one it was supposed to help.

Unbounded authority is worse, because the drift it produces is slow and invisible. Nobody notices that upper-funnel spend has been reallocated away until the pipeline stops growing, and by then the decisions that caused it are months of small, individually reasonable moves that nobody recorded.

A phase plan with stage-bounded limits and honest evidence periods is the arrangement that fits the cycle.

The most important limit in this version is that budget authority stops at the edge of a funnel stage.

Within demand capture, moving money between campaigns is a tactical decision that the person closest to the account should be making. Moving money out of awareness and into capture is a strategic decision with a two-quarter consequence, and it should require the person who owns the pipeline number to agree.

This is the specific protection this category needs. Capture always wins a short-horizon comparison, because the intent was already there, so a permission that treats the stages as interchangeable will empty the top of the funnel gradually and without a decision ever being made.

The evidence-to-advance column asks for something that has actually happened, and in this business most things worth demonstrating take a quarter.

A phase that ran for three weeks and produced a handful of leads has not shown that the authority is being used well; it has shown that nothing went badly wrong in three weeks. Advancing on that is advancing on comfort.

So phases here are longer than in other categories, and the evidence is usually a reconciliation rather than a result: two quarters where the ad account and the pipeline data agree, a full quarter of recorded changes with no reversal, a cohort that matured and was reported accurately.

Two things belong firmly in the approval column, and they are the two most likely to be changed quickly by somebody trying to help.

Changing what counts as an accepted lead changes every number the campaign is judged on, and it needs both marketing and sales to agree. A definition tightened quietly in one week makes the previous quarter look better and the current one worse, for no reason anybody can see in the data.

Changing routing — where leads go, who owns them, what fires when a form is submitted — is where campaigns quietly fail. A change made during a busy week, by somebody with good intentions, can lose a fortnight of leads without producing a single visible error.

Neither is a big change in effort terms, which is exactly why they need an approval rather than a limit.

Not everything has to advance at the pace of the cycle.

Creative rotation within an approved library is low risk and its effects are visible in engagement within days. Audience refinements inside a named segment are reversible. Both can reach a wider phase quickly, and doing so relieves most of the day-to-day friction without touching anything strategic.

The limit is the boundary of the segment and the library. A new segment is a new strategic decision; a new concept is a new brief.

Rules in this category need one extra condition: they may only act on signals that are actually readable in their timeframe.

A rule that reallocates on cost per accepted lead is defensible, if the window is long enough. A rule that reallocates on cost per form fill will optimise towards whatever produces the most unqualified leads, quickly and consistently, and it will look like it is working the entire time.

So the row for a rule should name the signal, the window, and what was tested against past data before it was given authority. A rule tested on two quarters of history is a rule somebody can defend; one built on last month is a guess with a schedule.

The reversal rule matters more here because the evidence arrives late.

When a quarter closes badly, the question of who was making decisions becomes charged, and a phase plan with a rule written in advance turns that into a mechanical step: the named condition happened, the area returns to approval, and the conversation is about what to do next rather than about whether somebody should have been trusted.

Write the trigger specifically, write who decides, and write it while the quarter is going well.

Seven checks. One row per area, no blanket permission. Budget authority stops at the edge of a funnel stage, and upper-funnel spend cannot be reduced under ordinary authority. Evidence periods are long enough for the cycle to fill them. Lead definitions and routing changes require approval. Creative and audience authority may advance faster, bounded by the approved library and the named segment. Automated rules name their signal and window and were tested against past data. And every change is recorded on the day, with its reason.