Annual Planning: Why Last Year's Plan Won't Work | SBI

Short answer: Most annual planning starts by rolling last year’s model forward: more reps for more bookings, more pipeline from the same channels, more output from new tools. Three of those equations changed. Growth now depends on productivity more than headcount, demand can’t be extrapolated from last year’s sources, and AI capacity isn’t the same as realized output. A plan built on the old math will reconcile in the spreadsheet and miss in the field.

Every planning season has a default move. Take last year’s actuals, apply a growth rate, and back into the headcount, pipeline and budget required. Finance checks that the model adds up. It does. The plan gets approved.

The problem isn’t the arithmetic. It’s that the relationships underneath the arithmetic have shifted, and most planning models haven’t caught up. As SBI put it in The Plan Is the Number, the financial model can reconcile while the commercial model does not.

Here are the three equations that changed, with what each one looks like in a real plan.

Equation 1: Does adding reps still add revenue?

The old equation: more sellers equals more bookings. If one ramped rep produces $700K, ten more reps produce $7M.

That math assumes every new rep is productive for the full year. They aren’t. SBI’s benchmark ramp time for B2B software sellers is 9 to 12 months. A rep hired in January on a nine-month ramp delivers roughly 60% of a full year’s output in their first year, and that’s if they start on January 1.

Run it through a plan:

That’s a $2.6M gap from one assumption. Add hiring delays and attrition, and it widens. SBI’s turnover benchmark is 15% to 20% a year, so some of those new reps won’t be there in December.

The alternative lever is productivity. Raising output 10% across an existing team of 40 ramped reps adds $2.8M with no ramp, no recruiting risk and no new fixed cost. Growth now depends on changing the productivity of the commercial system more than its size.

What to do: Model hiring with real ramp curves and real start dates. Then model a productivity scenario next to it and compare the cost and certainty of each.

Equation 2: Can you plan pipeline from last year’s demand?

The old equation: next year’s pipeline equals last year’s pipeline by source, plus growth.

That works when buyers behave the same way year to year. They don’t. More of the buying process happens before a buyer ever talks to a seller, through peers, analyst content and, increasingly, AI assistants. Forrester’s 2025 Buyers’ Journey Survey found 94% of business buyers now use AI during the purchasing process.

For planning, that means demand falls into three groups:

• Identified demand: buyers who have raised their hand through a form, an inbound call or a referral.
• Signal demand: accounts showing intent through research behavior, product usage or engagement, but not yet in conversation.
• Unidentified demand: buyers researching right now who haven’t shown up in any of your systems.

Most planning models only count the first group, because that’s what the CRM tracks. If identified demand is shrinking as a share of the market, extrapolating it forward overstates next year’s pipeline.

What to do: Break last year’s pipeline down by source and check which sources grew and which shrank. Then decide how the plan will find signal demand earlier, and assign that work to a named owner.

Equation 3: Does AI capacity become revenue?

The new equation everyone wants to use: AI saves each rep five hours a week, so capacity rises 12%, so bookings rise 12%.

Saved time is potential capacity. It becomes revenue only if it moves into selling activity and that activity converts at your current rates. SBI’s benchmark for healthy selling time is 60% to 63% of a seller’s week. If your reps are at 45% and AI frees up time that gets absorbed by more internal meetings, nothing changes in the number.

The same applies in marketing, RevOps and customer success. AI can produce more content, faster analysis and earlier churn signals. None of that shows up in bookings until someone redesigns the work around it.

What to do: For every AI productivity assumption in the plan, write down the mechanism: which hours, redirected to which activity, converting at what rate, measured how. If you can’t write the mechanism, don’t put the number in the plan.

What does this mean for next year’s annual operating plan?

Each equation produces the same failure pattern. The plan uses a relationship that held last year, the spreadsheet accepts it, and the gap only appears in Q2 when bookings fall behind.

The fix is to change the planning question. Instead of “How much more do we need?”, ask the four questions SBI uses to pressure test a commercial plan:

1. Where will the growth come from?
2. Where will the productivity come from?
3. What prevents us from delivering it today?
4. What must change?

If your plan can answer all four with specifics, it’s built on the new equations. If it answers them with growth rates applied to last year’s actuals, it’s built on the old ones.

Where should you start?

Take the three biggest assumptions in your draft plan, usually headcount, pipeline and productivity, and ask which equation each one depends on. Then check whether that equation still holds for your business.

For the step after that, read Which Number Is the Real Number? to see how to turn the board’s target into conditions you can test, or visit the SBI Annual Planning Hub.

The numbers will reconcile. The question is whether the commercial system underneath them can deliver.