Connected reading

This is Part 3 of Erin’s case.

If you’re starting here, begin with:

• Part 1 story: The Dashboard Said It Was Her Best Offer. It Was Quietly Draining the Business.

• Part 1 guide: Why Your Best-Looking Number May Be Lying to You

• Part 2 story: Knowing the Right Number Didn’t Fix the Business

• Part 2 guide: How to Build a Recovery Plan for a Small Business Using AI

For the practical companion to this article, read:

• How to Use AI to Review a Plan Before Drift Gets Expensive

Labor Day was for telling the truth

By Saturday afternoon of Labor Day weekend, Erin had been awake since 5:30.

Her daughter’s first game had started early. A second followed. By 11:00 the day’s work was done.

The next game was not until Sunday at 4:00.

That left a long stretch of dead time to fill in a team hotel off the interstate in one of the least exciting places she could imagine spending a holiday weekend.

Normally, that kind of time would have felt like punishment.

This time, it felt like an opening.

Her daughter’s best friend was on the same travel team, and the other girl’s parents offered to take both girls to a nearby mall for lunch and maybe a movie. Erin said yes before they finished the sentence.

She watched them leave laughing down the hallway, then went back into the room, pulled an antibacterial wipe from her bag, and wiped down the little hotel desk with more determination than it probably deserved.

Then she opened her laptop.

The hallway still smelled faintly like chlorine and industrial cleaner. The air conditioner ran loud and cold, then clicked off and let the room turn sticky again. Someone’s kids were stomping overhead. Somewhere nearby, a TV was on too loud.

It was not where she would have chosen to review the business.

It was where she had the time.

She sat down and, for the first time in weeks, asked the question she had been too busy to ask properly:

Is the plan actually working?

The answer, she already knew, was not going to be simple.

But before she got to what was not working, she had to admit something she had not been giving herself enough credit for:

the plan she built in June had already done real work.

No new retainers had been added. That alone changed the feel of the business.

The constant onboarding churn that used to eat half the day had almost disappeared. With time, clearer boundaries, and more repetition than she had hoped, the newer retainer clients had become less dependent on her and the team than they were in June. They still took more work than the legacy accounts. But they were no longer swallowing the day whole.

And the slow-paying clients she had tolerated for too long had finally been dealt with.

The Monday after she built the June plan, she had gone back to ChatGPT and asked for help drafting the message she had been avoiding. Not a threatening message. Not performative. Just clear.

Pay on time, or this relationship will not continue.

She had expected drama.

There was less of it than she thought.

One client apologized and got current. One pushed back and then paid anyway. One dragged their feet just long enough to prove they were not worth saving.

The plan had not fixed the business.

But it had stopped the first wound from getting deeper.

That mattered.

It mattered enough that she and the team finally had the breathing room to push the healthier strategy they had chosen.

Which was why what came next bothered her so much.

Because they had worked the plan.

And it still was not moving fast enough.

When Erin opened the July-and-August numbers, the problem was not hard to see.

The business was healthier than it had been in June.

It was also not yet healthy enough.

Erin’s Summer Plan vs. Actual

Reviewed Labor Day weekend

She stared at the table longer than she wanted to.

The direction still looked right. That was the frustrating part. They had stopped the bad growth. They had pushed strategy first. They had not drifted back into the retainer habit. And yet the business was not moving through the new plan with enough speed to feel stable.

This was the kind of moment that makes an owner do something dumb.

Throw out the strategy.

Panic back to whatever used to feel familiar.

Convince herself that the old way, even if it was worse, at least “worked.”

But the numbers did not say the strategy was wrong.

They said something narrower and more uncomfortable:

the plan was directionally right, but the business was not moving through it fast enough.

That was different.

And it needed a different response.

So instead of jumping to conclusions, she opened the sales process.

That was where the truth got louder.

Where the sales process was getting stuck

Strategy and planning work — July and August

Additional signal: 4 of the 9 full strategy / sales conversations had been rescheduled at least once. 3 of those had been rescheduled more than once.

By then, she could finally see where things were getting stuck.

Erin did not mainly have a lead problem. The top of the funnel was fine. People were interested. Conversations were happening. Enough warm prospects were entering the process that, six weeks earlier, she would have taken it as proof the strategy was working.

But good conversations were not turning into signed work fast enough.

That was the real drag.

She had mistaken activity for movement.

And once she saw that, another realization followed right behind it:

some of the truth had already been sitting there in her notes and follow-up emails.

She just had not been reading it closely enough.

She opened the folder where she kept notes from prospect calls.

Then the email threads.

The pattern was not subtle once she was willing to look at it plainly.

What prospects were actually saying

• “This sounds good. Let’s revisit after Labor Day when our GM is back.”

• “I need my partner on the call before we can move forward.”

• “We’re interested, but this feels bigger than what we need right now.”

• “Can you scope something smaller to start?”

• “The direction makes sense. Timing is the hard part.”

None of that sounded like rejection.

It sounded like people who were interested but not quite ready to move.

Some of it was just the normal drag of late summer.

Some of it was a sign that the offer, as she had packaged it, was still a little too big or too slow for the moment they were in.

That was the part she had not seen clearly enough.

So she opened ChatGPT again.

Not for a new master plan.

For a review.

That was the shift.

In June, she had asked AI to help her build a recovery plan.

In September, she was asking it to help her look at the plan honestly and figure out where it was getting stuck.

She already knew what the numbers were telling her.

She just needed help saying it clearly before she decided what to do next.

I need help reviewing my plan against reality.

I do not think the strategy is wrong, but I do think part of it is getting stuck.

Here was the plan: [PASTE]

Here is what actually happened: [PASTE]

Here are my notes and emails from prospects who showed interest but did not sign: [PASTE]

Help me do five things:

1. tell me where the plan seems to be getting stuck

2. help me see what really matters here and what is just the normal mess of doing business

3. help me identify the real bottleneck now

4. suggest 2–3 adjustments that fit the business I’m actually trying to build

5. help me choose one change to make for the next 30 days without throwing out the whole strategy

Keep this practical and in plain English.

The answer came back faster than it had in June but more carefully than the old growth-plan nonsense from winter. That was becoming familiar. The more real she was with AI, the more useful it got.

The diagnosis was not dramatic.

It was plain.

People were interested.

The problem was that too few of those conversations were turning into signed work fast enough.

The holdup was happening after the interest and before the yes.

The repeated rescheduling pointed to a mix of things:

• the time of year

• slow internal decision-making

• and an offer that may have felt bigger than some prospects were ready to move on quickly

Then came the sentence Erin read twice:

You have enough here that notes alone may not be enough. If you record and transcribe these calls, with permission, I can help you spot the patterns more clearly — where people are hesitating, what they are not saying directly, and what they may actually be asking for underneath the surface.

She leaned back.

That had not occurred to her.

Not because it was impossible.

Because she had still been treating AI like something you open when it is time to sit down and “do strategy,” not something you bring into the work while it is actually unfolding.

That suggestion changed something.

Because it showed her a different kind of partnership.

Not just:
make the plan, then come back later.

But:
keep bringing the real work into the room while it is still happening.

That was new.

And it made the rest of the answer land harder.

AI helped her sort out what really mattered and what was just the normal mess of a business moving through late summer.

Some of it was ordinary:

• some slowdown this time of year was normal

• some lag between strategy and cash was expected

• some website work was always going to take time to complete

But some of it was telling her something useful:

• the strategy itself was still the right direction

• the current version of the offer was too heavy for part of the market in that season

• too many prospects needed a smaller or faster first step

• and she needed to review the plan more often than once a quarter if she wanted to catch this kind of drift before it got expensive

Then it suggested three adjustments.

One stood out immediately.

Do not throw out strategy.

Package it differently.

A smaller, faster-close version of the same high-value thinking.

A lighter first step for buyers who could not move on the full offer quickly — especially at a time of year when decision-makers were away and companies were not moving cleanly.

Then it gave her another idea she probably would not have reached as quickly on her own:

if the time of year was part of the problem, use it.

Look for businesses entering a planning season now.

Retailers getting serious about holiday execution. Winter-driven businesses heading into their real season. Companies that could move faster because the need was about to become urgent.

That was not a new business.

It was the same strength, aimed at a moment that fit it better.

The plan was not collapsing.

It was telling her where it needed to bend.

So she spent the next half hour with GPT building the revised move.

Erin’s Revised Fall Plan

September–November

This was not the kind of revision that makes a founder feel heroic.

It was better than that.

It was useful.

It kept the strategy, changed the packaging, and used what she had learned instead of pretending she had not learned it.

And it made one other thing painfully obvious: a quarterly review was too slow.

The plan had not failed.

It had drifted.

And if she had been reviewing the right things every week — with real attention, not just a glance at a dashboard between meetings — she would have seen it sooner.

That might have been the easiest lesson in the whole case. And maybe the one most founders ignore.

So she built herself a weekly review page. Not a giant KPI dashboard. Just enough to stay honest.

Each week, she would check:

• strategy / sales conversations completed

• conversion from strategy conversation to signed work

• cash collected from healthier work

• stalled prospects older than 14 days

• new work sold by healthy offer type

The point was not to drown in numbers.

The point was to stay close enough to reality that the next drift did not get expensive before she noticed it.

By the time she closed the laptop, the frustration she had started the afternoon with was gone.

Not because the business was fixed.

Because she understood it more clearly than she had four hours earlier, and she had a version of the plan that was more honest about what she was actually dealing with.

The plan had not worked exactly the way she hoped.

But it had worked enough to reveal the next truth.

And now she had a better version.

That was what good planning actually looked like inside a real business.

Not getting it right once.

Getting it true enough that reality could keep improving it.

She no longer needed the plan to be right in every detail.

She needed it to keep getting truer.

She got a text just then.

The team dinner had been decided. Hibachi. Thirty minutes.

Erin smiled, shut the laptop, and stood up. She packed her bag, glanced in the mirror, and cleaned herself up just enough to look like she had not spent the better part of the afternoon sitting on a hotel bed with a spreadsheet and a strategy problem.

Then she grabbed her keys and stopped long enough to pull her rain jacket off the chair.

She was not about to ruin another decent shirt with an ill-flipped shrimp to the chest.

On the walk out, that was the part she finally understood about working with AI.

Part 1 taught her that the dashboard had been answering the wrong question.

Part 2 taught her how to build a plan honest enough to test.

Part 3 taught her something harder: a plan is only useful if you stay close enough to reality to adjust it before the drift gets expensive.

The real shift came when she stopped treating AI like something you use once to get a big answer and started bringing it into the work as it was actually happening.

The first use gives you a plan.

The second gives you a way to keep getting it right.

AI could help her see more clearly.

But she still had to decide what to do with what it showed her.

That was still her job.

Continue Erin’s case

• Go back to Part 1 story: The Dashboard Said It Was Her Best Offer. It Was Quietly Draining the Business.

• Review the Part 1 guide: Why Your Best-Looking Number May Be Lying to You

• Go back to Part 2 story: Knowing the Right Number Didn’t Fix the Business

• Read the Part 2 guide: How to Build a Recovery Plan for a Small Business Using AI

• Read the companion guide to this article: How to Use AI to Review a Plan Before Drift Gets Expensive