AI killed AOPs. Or that’s how you’re probably using it today

Annual Operating Plans are still useful, even with AI productivity boost

Stop writing plans on vibes. Measure the lift, then write the plan.

The founder question that exposes the real problem

A founder asked me something I keep hearing in different forms.

• “How am I supposed to build an AOP when I have no idea what AI will replace or automate?”

That question sounds tactical.

It is actually structural.

Because an AOP is a commitment.

It is a set of numbers you put in front of a board, a team, and yourself.

It is supposed to be grounded in capacity, conversion, and constraints.

AI, as most teams are using it, is not grounded in any of that.

It is a productivity rumor.

When you build an AOP on a rumor, you do not get a plan.

You get a story that someone will treat like a contract.

Why this is getting worse, not better

Two forces are colliding.

• First, founders are watching large companies do layoffs “because AI” and assume similar lift is available in a smaller org.

• Second, boards still want certainty. The calendar still turns. Budgets still get set. Headcount still gets approved or denied.

So AI becomes the bridge people use to connect aggressive targets to an unchanged operating system.

“We will grow faster next year.”

“How?”

“AI.”

That is not a strategy.

That is a placeholder.

And placeholders are dangerous because they look finished on a slide.

The rule: if AI is not in the workflow, you are guessing

If you have not tried AI inside your business, you are starting from scratch.

You will not be able to see where productivity can be picked up because you do not have baselines.

You also will not know what breaks when you scale usage across the team.

If you have tried AI, the opposite problem happens.

Teams feel momentum and assume it equals measurable lift.

It might.

But the only honest move is to quantify it.

No board deck should contain “AI will fill the gap” unless you can point to the gap and prove the fill.

The fix: create the AI Productivity Metric

• You do not need a year of data.

• You do not need a complex model.

• You need a few months of clean, comparable tracking.

Three to six months is enough to create an AI Productivity Metric.

Pick 2 or 3 measures tied to revenue, throughput, or cycle time.

Measure before and after AI became normal in the workflow.

Use the delta to inform next year’s operating assumptions.

Examples that tend to matter in GTM and ops:

Speed to lead

• Time from inbound to first human touch, then to first meeting booked

Sustainable daily outreach

• Not the best day. The median day, per rep, with quality held steady

Research depth per account

• How many relevant insights make it into the first message or call prep

Stage conversion rates

• Meeting to qualified, qualified to pipeline, pipeline to close

Cycle time on operational work

• Ticket time, lead routing time, enrichment time, quote creation time, QA time

You are looking for proof of uplift.

Not anecdotes.

Not “it feels faster.”

A simple way to compute it:

• Choose your 2 or 3 measures

• Normalize each to a 0 to 100 index against your pre AI baseline

• Average them

• Track it monthly

That is your AI Productivity Metric.

Then flip it. The inverse is your Gap for AI

Once you have the AI Productivity Metric, you can stop guessing what AI “should” do.

Now you can ask a sharper question.

What lift did we actually capture, and what lift are we assuming next year?

The inverse is what I call the Gap for AI.

It is the portion of your AOP that depends on productivity lift you have not proven yet.

If your plan requires a 25 percent productivity increase and your metric shows 6 percent, your gap is 19 percent.

That gap is not “free.”

It has to come from somewhere:

• More pipeline

• Higher conversion

• Higher ACV

• Higher retention

• More headcount

• Less churn

• Less waste

• Better segmentation

• Better product packaging

If you cannot name where it comes from, it is not a plan.

It is hope with formatting.

The trap: confusing last year’s growth with next year’s probability

I recently spoke with a founder who had roughly 80 percent YoY growth.

The company was under $5M.

They wanted to do it again.

Is it possible?

Sure.

Is it likely, purely on probability?

No.

As companies scale, the base gets bigger.

Markets get tighter.

Competition wakes up.

Channels saturate.

The “easy” wins get used up.

When I asked how they would repeat the growth, the answer was “AI has been a huge help.”

I asked the only question that matters.

“What did AI change, specifically, and by how much?”

There were no numbers.

No baseline.

No delta.

So the plan was being written as if AI would deliver the missing capacity.

And once that plan gets approved, the expectation becomes real.

Now the board expects outcomes that might be physically impossible with the current system.

Human oversight is still part of the cost model

There is another subtle problem hiding inside the AI story.

Every AI tool includes some version of “it might make mistakes.”

That is not legal fluff.

That is an operating constraint.

If the tool makes mistakes, then humans must review outputs.

Review time is labor.

Labor is capacity.

Capacity belongs in the AOP.

If you assume automation without accounting for oversight, you will undercount cost and overcount throughput.

That creates the worst kind of miss.

A plan that looked efficient on paper, then collapses under reality.

The goal is not to avoid AI.

The goal is to treat it like any other system change.

Measure the lift, price in the constraints, then write the plan.

How to use this in your next AOP cycle

Here is the practical sequence I recommend.

1. Pick 2 to 3 metrics that tie directly to revenue throughput or operating cycle time

2. Establish a pre AI baseline, even if it is messy

3. Track the same metrics for 3 to 6 months after AI adoption becomes consistent

4. Compute your AI Productivity Metric and trend it monthly

5. Identify the Gap for AI in your draft plan

6. Replace that gap with explicit levers, or shrink the plan

This is not about pessimism.

It is about not signing your name to numbers that depend on invisible productivity.

AI did not kill AOPs.

Vibes killed AOPs.

If you want to use AI in the plan, earn it with measurement.

Then you can talk about productivity with a straight face.

And your board deck stops being fiction.