Daniel J. Jacobs 🇬🇧

Middle managers aren’t the problem the systems trapping us are. I wrote The Strategy Bridge to change that: practical tools to think clearly under pressure, stop regretting decisions, and actually connect the systems that don’t talk. No budget, no sponsor, no transformation program required. mybook.to/strategybridge What’s the biggest system pain you’re dealing with right now?

Check out the latest article in my newsletter: Pausing modernisation to fund AI isn't a saving. It's a loan against the AI's return. linkedin.com/pulse/pausing-m… via @LinkedIn

A business running above its sector's IT spend range isn't automatically wrong. Investing deliberately, into a real capability build, has a defensible story. Sitting above range with no story is a governance gap, not a strategy.

Shadow SaaS. Orphaned licences. Duplicate tooling. These have been quietly compounding IT budgets since long before generative AI arrived. The newest category is just the most visible current example of an old problem.

Itemise the current IT budget into benchmarked categories before the next cycle starts, not during it. That single exercise turns next year's renewal from a negotiation under pressure into a scheduled review.

Only 31% of organisations have accurate visibility into their AI software spend. Only 36% have complete visibility across the whole IT estate. (Flexera, 2026 State of ITAM Report.) AI didn't create this gap. It's exposing one that predates it by a decade.

IT cost optimisation is not a cutting exercise. It's an ownership exercise. You can't negotiate your way to a well-run budget once a year under time pressure. You build a category structure with named owners instead

Most "IT cost optimisation" is a vendor renegotiation exercise run once a year under time pressure. That produces a one-off saving and leaves the underlying spend structure untouched for next year's scramble.

The average IT budget sits at 5.7% of revenue (Computer Economics / Gartner IT Key Metrics / IDC, 2026). Manufacturing runs closer to 2%. Financial services and tech run 7-10%. Most boards don't know which range they're in.

Gartner revised its 2026 IT spending forecast upward three times this year: 9.8% -> 10.8% -> 13.5% growth. Total now $6.31T. Data centre systems +55.8%. Software +15.1%. Most boards' approved budgets moved zero times in response.

Ask your board this: can anyone explain a specific IT budget line's variance without escalating to the CIO first? If not, you have one signature under one number, not a budget you can manage.

If a confident voice rang your IT helpdesk at 3 pm on a Tuesday and asked to reset a privileged password, what would actually happen? For M&S, the answer cost around £300m. Why the helpdesk is the new perimeter, and the fix that isn't more training: open.substack.com/pub/starkh…

Middle managers aren’t the problem. The broken systems trapping us are. I wrote The Strategy Bridge to fix that: practical tools to think clearly under pressure, stop regretting decisions, and actually connect the systems that don’t talk. No budget. No sponsor. No transformation program required. → mybook.to/strategybridge What’s the biggest system pain you’re dealing with right now? #MiddleManagement #Leadership #Strategy

8.5 million Americans hold multiple jobs. Dad pulled over after a full day no sleep to provide for his family, neighbors called cops instead of knocking. Officer listened without the ID hassle and handed over his coffee. Actual service.

Here is how to know if you are the problem. Your direct reports cannot answer a substantive question without first checking with you. They stopped trying about six months ago. You interpreted their silence as alignment. It wasn't. Organisational silence has a precise definition in management research. Employees withhold concerns, ideas, and pushback because they've learned the cost of speaking exceeds the benefit. The cost is rarely dramatic. It's subtler. The project that keeps failing is being rebooted and is failing again. A phoenix nobody is allowed to kill. The same faces in every meeting, regardless of the agenda or who actually owns the decision. Nobody brings problems upstairs. Only solutions that were pre-cleared before the meeting started. ◆ Gallup's research across 35 million employees found managers account for 70% of the variance in team engagement. Not strategy. Not pay. ◆ DDI's Global Leadership Forecast found 57% of employees have left a role specifically because of their manager. Most gave a different reason on the exit form. ◆ Research into executive behaviour found that controlling leadership at the top restructures decision-making three to four layers below. Nobody acts near the waterline until the signal from the bridge is clear. They read the quiet room as a sign of cohesion. A new arrival on day 30 sees it differently. When nobody challenges the leader in the room, that's not a high-performing team. That's a team that has quietly calculated the risk. When the captain steers every rudder, the ship doesn't move faster. It learns to wait. Silence is not agreement. Compliance is not confidence. Deference is not trust. What's the earliest sign you've seen that a team had quietly stopped managing the business and started managing the boss instead? #Leadership #Management #FractionalCIO #ExecutiveLeadership #OrganisationalCulture

Fast and wrong costs more than slow and right. Most organisations are learning this the hard way. Completing a task and solving the underlying problem are different activities. The human brain treats them identically. This is not a discipline gap. It is neuroscience. Closing a task triggers the same dopamine response whether the problem was solved or not, creating a self-reinforcing loop where teams seek closeable tasks for the biochemical reward. Development teams spend 30-50% of their time on rework, fixing code that was already marked done. The Stripe Developer Coefficient quantifies this at 42% of the average working week. The US annual cost of poor software quality: $2.41 trillion. (CISQ, 2022.) Goodhart's Law is operating in every sprint. When velocity becomes the target, it stops being a useful measure. Developers inflate estimates, teams split stories, and the metric improves while the product does not. On 1 August 2012, Knight Capital deployed an update to 7 of 8 production servers. The eighth contained dormant 2003 code that nobody had removed. Within 45 minutes: 4 million erroneous trades, $7 billion in unwanted positions, a $440 million loss. Nobody decided the eighth server was acceptable. It was just not the ticket anyone had opened. On 19 July 2024, CrowdStrike bypassed standard validation for a routine content update. 8.5 million Windows devices crashed simultaneously, with no staged rollout and no circuit breaker. Global losses exceeded $10 billion. Someone applied the patch. Nobody removed the original risk. DORA's research across 32,000 professionals found high performers deploy 208 times more frequently than low performers, with change failure rates below 15%. The lowest performers fail on nearly half their deployments. Speed and quality compound together or erode together. Three signals your team is optimising for closure, not outcomes: ◆ Stories split before refinement to generate more done events, not because the problem decomposed. ◆ P1 labels capture immediate engineering attention that architectural debt never will. Velocity climbs. Customer outcomes do not. Applied the patch. Closed the ticket. The system failed three weeks later. What is the most dangerous thing your team has ever marked done?

Satya: "AI's biggest obstacle is changing how people work." Translation: we built the tools before the organisations were ready. Every enterprise AI failure I've seen follows the same sequence: Buy the platform Skip the process redesign Blame adoption The technology works. The operating model doesn't. And no amount of Copilot licences fixes a leadership problem.