Russ Reeder is CEO of KeyDelta. Nine PE-backed CEO and COO seats; 50+ acquisitions. We fix the operating model before AI goes in.

When I was a kid, my grandfather gave me one piece of advice about books: read autobiographies. If they tell the truth, you’ll learn something.

That advice has stayed with me through 30 years of running technology companies. A story, a strategy or a product plan is only useful if it’s honest about what doesn’t work. And the person least able to see what doesn’t work is usually the one who built it.

I’m living that right now. This month I finished the manuscript for a book on leadership. Before it goes into production, I’m sending it to a small group of trusted readers: close friends, mentors and leaders I respect. After this many drafts, I know what I meant. I need to know what a reader gets.

The same step belongs before every launch a leader makes: a upcoming strategy, a product, a reorganization or an AI pilot. Many leaders skip it. Or worse, they ask for approval and think they asked for the truth.

According to recent reports, when a CEO presents a proposal, the room tends to nod. That isn’t always agreement. Often it’s the safest thing to do in front of the person who controls budgets and careers.

Sources indicate that nokia shows how expensive that can get. INSEAD’s Quy Huy and Aalto University’s Timo Vuori interviewed 76 managers, engineers and outside experts and found that fear shaped how information moved inside the company. Some middle managers stayed silent or passed up filtered, optimistic information, while top managers pushed for faster delivery on an incomplete picture. Everyone shared the ambition. The facts got softer at every level.

A head nod is not a commitment. If the only feedback you’ve collected came from people who report to you in a meeting you ran, you’ve collected approval.​.

It has been revealed that I learned this the hard way. At a managed services organization I led, our AI engineers built a system that automatically filled in support tickets. Months later, the support reps still weren’t using it. In a weekly standup, the engineers asked why.

The head of customer support answered: “I don’t know why they’re not using it. I haven’t asked them.”.

Recent findings suggest that the call went silent. We built the tool around how managers described the work, not how reps did it or how the work should be done. Once we asked, we had to start almost from scratch. We rebuilt the second version around the outcome the customer needed, and it could resolve certain issues without anyone touching the ticket.

Industry observers note that the truth was available the whole time, but nobody went to get it. Your trusted readers should include the people who will use, deliver or be measured on what you’re launching. Their view is the one a leadership meeting rarely hears.​.

“What do you think?” invites politeness. Specific questions invite the truth. Here are the three I’m asking my readers. I believe they also work for almost any launch:.

1. Where did you stop believing it? This may be where a claim felt too clean, a number too convenient or a timeline too optimistic.

Recent findings suggest that 2. Where did you see your own work in it? That tells you which parts are landing and which are abstract.

Industry observers note that 3. Where did you lose interest? Every strategy has a section people skim. Find it before your customers or your board do.​.

It has been revealed that ​Then go first. Tell them what you’re worried about. In my experience, people are far more willing to point out a problem after the leader has named one.

Feedback that changes nothing teaches people to stop giving it. Before you ask, settle three things:.

It has been revealed that 1. Who owns the changes. This should be one name, rather than “the team.”.

It has been revealed that 2. When the review closes. Aim for a specific date so the feedback leads to a decision instead of an open-ended debate.

It has been revealed that 3. What would change the plan. Write down what you would need to hear to delay, redesign or stop. If nothing could change it, then don’t ask. You’re not asking for feedback. You’re asking for applause.

Sources indicate that when you get answers you don’t like, thank the person who gave them. The first honest critic sets the price for the next one. If candor costs them something, it will be the last honest answer you get.​.

This step comes before production for a reason. A sentence is easy to change in a manuscript and expensive to change in print. A process is easy to fix on a whiteboard and expensive to fix after it’s built into software. A strategy is easy to adjust before you’ve funded it and expensive to unwind once teams are executing it.

Sources indicate that the truth costs the least before you launch. After that, your customers will tell you anyway, and they won’t be as kind. Find the people who will tell you where they stopped believing you, and ask them while you can still use what they say.​​​.

Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?

→ Read original article