The AI Implementation Graveyard: Why 73% of Projects Fail Before Launch

Howdy, I was going to post about the risks and dangers of AI that we need to be aware of. That’s not a very positive story for start of the year, so will publish that in future.

If you’re enjoying this newsletter, please share it! Forward this to a friend/colleague and have them subscribe at aisecondact.com/subscribe. Thanks!

👉 You can also just hit "Reply" and let me know what you want more (or less) of, or use the poll to give feedback.

My goal is to make this as valuable and practical as possible as we navigate the new AI era. 🚀

Here are a few things I found recently:

The vendor was smooth. The demo was slick. Eight months later, the "transformational" project was quietly shelved.

RAND Corporation's analysis confirms that over 80% of AI projects fail—twice the failure rate of non-AI technology projects. WorkOS

MIT's research found that only about 5% of AI pilot programs achieve rapid revenue acceleration; the vast majority stall, delivering little to no measurable impact on P&L. Fortune

S&P Global found that the share of companies abandoning most of their AI initiatives jumped to 42% in 2025, up from 17% the previous year. The average organization scrapped 46% of AI proof-of-concepts before they reached production. CIO Dive

The vendor pitch always sounds the same: Better data. Smarter decisions. Competitive advantage.

The reality is messier and here's what actually kills them:

1. The "Solution Looking for a Problem" Disease

Most AI projects start with technology, not pain. Someone attends a conference, sees a compelling demo, and returns determined to "do AI."

The question that never gets asked: What specific, expensive problem are we solving?

McKinsey's 2025 AI survey confirms this pattern: organizations reporting "significant" financial returns are twice as likely to have redesigned end-to-end workflows before selecting modeling techniques. WorkOS

Translation: The winners start with the problem, not the tool.

2. The Data Quality Time Bomb

Every vendor assumes your data is ready. It's not.

Informatica's CDO Insights 2025 survey identifies the top obstacles to AI success: data quality and readiness (43%), lack of technical maturity (43%), and shortage of skills (35%). WorkOS

The old maxim that 80% of machine learning work is data preparation? Still true. Maybe more true.

3. The Governance Vacuum

Legal wants guardrails. IT wants security. Operations wants reliability. Leadership wants results yesterday. And nobody's talking to each other.

Companies cited cost, data privacy and security risks as the top obstacles to AI success. CIO Dive

Without a clear governance framework before you start, you're building on sand.

4. The Pilot Purgatory Problem

Gartner found that only 48% of AI projects make it into production, and it takes 8 months on average to go from AI prototype to production. Informatica

Pilots are comfortable. Production is scary. So projects linger in that cozy proof-of-concept phase until budget reviews kill them or sponsors move on.

Before you give the ‘go’ to any AI initiative:

Question 1: "What would we do without AI?"

If you can't clearly articulate the current state cost (in dollars, time, or risk), you don't have a business case. You have a science fair project.

Question 2: "Who decides if this succeeds?"

Not who sponsors it. Not who builds it. Who will look at results in 6 months and make the call on whether it's working?

Question 3: "What breaks if this works perfectly?"

This is the uncomfortable one. If your AI actually delivers, what processes change? What roles shift? What sacred cows get slaughtered?

I've turned these questions into a one-page pre-mortem checklist I run before any AI initiative crosses my desk. Three questions. Two minutes. Saves months of pain.

Want it? Reply to this email with "PRE-MORTEM" and I'll send it over.

Here's what the vendors won't tell you: AI success isn't about the model. It's about the machinery around it—the data pipelines, the governance frameworks, the change management, the executive air cover.

Winning programs invert typical spending ratios, earmarking 50-70% of the timeline and budget for data readiness—extraction, normalization, governance metadata, quality dashboards, and retention controls. WorkOS

The organizations that clear the failure statistics start with unambiguous business pain. They invest disproportionately in trustworthy data. They build human oversight as a feature, not an emergency valve. And they treat AI deployments as living products with success metrics tied to real dollars.

Weekly AI strategies for operating executives
— Brett

👉 Hit “Reply” and share your experience — I read every one!

_{Picture by Scott Rodgerson on Unsplash.}