95% of custom AI initiatives deliver zero measurable return.

Not "disappointing return." Zero.

That's the number Andrea Hill opened with on The Sales Show this week, and it hit hard.

Because most founders I talk to think they're in that 5%. They're optimistic. They've bought the tools. They have a plan.

But the data says otherwise.

Here's the difference between the 95% that fail and the 5% that don't.

THE TRAP: STARTING TOO BIG

The number one reason AI projects fail is simple: companies start too big.

They want to transform five workflows at once. They hire consultants to build custom platforms. They spend $100K before a single real user touches the system.

Massive AI implementations are a trap.

The projects that are returning real investment? Under $15,000. Often under $5,000. Gated with a clear time limit and a defined success metric.

$15K and under. That's not a starting budget. That's the final budget for a whole first project.

THE FIX: ONE WORKFLOW

"One workflow at a time," is what I said back to Andrea.

Not "one tool." One end-to-end business activity.

For example: the journey from stranger to booked call. From initial conversation to signed offer. One of those workflows, nothing else.

The constraint forces clarity. It also forces success...

because the scope is small enough that you can actually measure whether it worked.

THE STEP NOBODY TAKES

Before you touch a single AI tool, Andrea prescribes something called a friction mapping exercise.

You map the workflow you want to automate. You find where things slow down. Where the ball gets dropped. Where prospects should be moving faster but they're getting stuck.

You fix that first.

"If your process is fundamentally broken or if your data is disorganized, then adding AI just helps you do the wrong thing faster."

You can't automate your way out of a bad process.

WHERE THE REAL ROI IS

Andrea's clients see the fastest ROI in two places.

First: CRM and pipeline management. Most small business owners are doing this manually, running reports from memory, following up when they remember. An AI agent that handles the pre-call research, the follow-up timing, and the offer tracking can cut hours every week immediately.

Second: the offer-sending workflow. Simon and Andrea both confirmed this on the call: the gap between "ready to send the offer" and "offer signed" is where most deals die. An AI-powered offer-sending agent.. one that personalizes, follows up, and surfaces the right context.. has been one of the fastest ROI tools in their combined client base.

THE FREELANCER WHO BUILT IN THREE WEEKS

Here's the example that made the scale of this concrete.

Simon's audiobook. A professional audiobook would have cost $50,000 to produce.

With AI... it cost $200 a month. And it's being recorded in three weeks.

$50,000 vs. $200/month.

That's not a marginal efficiency gain. That's a structural change. in what's possible for a founder who works alone.

Scale that thinking to your B2B software team. Half the code is now written by AI. One-third of the engineering headcount doing the same output. Or double the output with the same team.

THE RULE

Start small. Fix the process first. Then automate one workflow. Measure it in 30 days.

If you can't show measurable improvement in one metric within a month, you implemented the wrong thing first.

Not the biggest thing. Not the hardest thing. The wrong thing.

Pick something painful. Something you're already doing daily. Something with a clear before/after metric.

Then let AI handle the low-cognitive-load parts of it.. the prep, the research, the follow-up, the admin.. and free yourself up for the judgment calls that only you can make.

That's the 5%.

Most founders are looking for ROI in the wrong place. 

They buy the tool. They set up the workflow. They announce to the team.. 'We're using AI now.'

And then.. nothing moves.

Revenue stays flat. The team goes back to doing things the old way. Six months later the founder wonders why AI didn't work for them.

Andrea Hill has seen this pattern dozens of times. And she knows exactly why it happens.

'The problem is not the AI. The problem is that founders try to automate activities instead of automating decisions.'

That's the insight that changes everything.

AUTOMATION VS. DECISION SUPPORT

Most AI implementations fail because they try to replace human activity. Sending more emails. Writing more content. Running more reports.

But the real ROI in AI comes from something different. It comes from better decisions.. faster.

Which leads to close? Which follow-up actually moves the deal? Which segment is about to churn?

When AI informs those decisions.. everything else accelerates.

THE 3-STEP FRAMEWORK

Andrea walks every client through the same three steps:

Step 1. Map your highest-stakes decision points.
Not 'what tasks can AI do?' but 'where do we make decisions that cost us money when we get them wrong?'

Step 2. Feed those decision points with data.
Most founders are making $50,000 decisions on gut feel and a CRM 60 days out of date. AI can fix that.. but only if data inputs are right first.

Step 3. Run one loop, not ten tools.
The founders who see ROI fastest build one feedback loop and let it compound. Not ten AI tools that each do something slightly useful.

THE 30-DAY ROI TEST

If you can't see measurable improvement in one revenue metric within 30 days.. you've implemented the wrong thing first.

Not the biggest thing. Not the most complex thing. The wrong thing.

The fastest ROI comes from implementing AI where the decision is already being made.. and just making it sharper.

Start with the conversation after the demo. Most B2B founders have no system for what happens after a prospect says 'interesting.. send me more.'

That gap is where deals die.

AI can close that gap in 48 hours.. if you know what to feed it

Watch the full episode with Andrea Hill here: https://youtu.be/7EhB44VXiDY

Sprint Club members get the full implementation guide: https://www.strategysprints.com

keep rolling, Simon & The Sprinters