Stop The Bleed
Contractors filling gaps that should not exist. Infrastructure sitting idle. Vendor contracts renewing without scrutiny. Here is how I built the evidence, made the cases, and converted waste into savings without touching headcount.
The Office Space problem
The situation
The pattern repeated across every organization. Budget going to specialist contractors because nobody had built the internal capability. Infrastructure outliving its purpose with no one willing to make the call. Vendor contracts auto-renewing because challenging them required evidence that had never been gathered.
None of it looked reckless from the inside. Every contractor, every idle server, every renewal had a reason the day it was signed off. What was missing was anyone going back to ask whether that reason still held. Cost rarely creeps in through dramatic decisions. It accumulates one defensible line item at a time, until the total is too big to ignore and too tangled to unpick.
Each of these problems has the same root: nobody had visibility over what was being used, what it cost, and what it was actually delivering. Without that picture, every line item looks justified to whoever signs it off.
Patterns found across engagements
What gets signed off
Actually used?
never measured
What it really costs?
never totaled
Delivering value?
never proven
What nobody can see
The approach
Replacing external dependency by closing the capability gap
External contractors and development partners stay on the books for one of two reasons: the work genuinely requires specialist capacity, or the internal team was never built to cover it. The second case is a structural problem dressed up as a staffing one, which makes the fix a capability decision, not a headcount one.
"We need more headcount"
A staffing problem on the surface.
We need internal capability
A capability decision underneath.
01 · IDENTIFY
Map the gap
What the internal team couldn't yet cover, and why.
02 · BUILD
Close the gap
Cross-training, process, tooling, before removing anything.
03 · CASE
Build the case
Current cost, what the team could own, risk-rated view.
04 · REMOVE
Dependency out
Once capability threshold was documented and verified.
At the start there was no internal capability in the specialist middleware platform, so contractors were brought in. They were used deliberately, not just to fill seats. They stood the platform up, wrote the documentation that did not exist, trained the offshore team, and ran hiring and interviews. In parallel, a T-shaped cross-training program built a shared internal team that could carry the work on its own.
Once that capability was documented and verified (the four steps in the diagram above), the three contractor/partner specialists came off the books. That removed $27k/month in external spend. The dependency had been engineered out, not cut and hoped for.
The team-building side of this story
The cross-training and hiring that built the internal team is the harder half of this. How a 30-person middleware team went from worst performing to best is the full account.
Read Build What Lasts →The same instinct, applied to AI
Building a capability in-house instead of renting it shows up again with AI: a triage layer the team owns and maintains, with cost controlled on every call, instead of a vendor bot nobody can steer.
Read The Pivot Point →Decommissioning infrastructure nobody wanted to touch
Legacy infrastructure persists because decommissioning creates risk and nobody wants to own it. The default position in every organization is to leave it running until someone else makes the call.
I cross-referenced monitoring data against the CMDB to build a clear picture of what was active, dormant, and genuinely idle. Where the data alone was ambiguous, I went directly to the teams responsible and had the conversation in the room. "We might still need that" does not survive a direct question backed by six months of zero-transaction logs.
Infrastructure Inventory
Monitoring data cross-referenced against the CMDB.
"We might still need that" doesn't survive six months of zero-transaction logs.
This was won one system at a time. Meeting after meeting: reviewing incident and change history, pulling logs, connection data and packet captures, and chasing the long-tenured owners who "remembered" a dependency. For each system I built an evidence file of logs, emails and monitoring, until the "we still need it" claim collapsed under it. Some of the pushback turned out to be miscommunication or simply false reporting. Where the data was clean, a low-risk migration made the removal painless: a server upgrade and a DNS swap, zero impact.
The decommission case was approved. $150,000 in annualized infrastructure costs eliminated.
How the governance model made this possible
The infrastructure decommission did not happen in isolation. The same vendor and change-governance rebuild made the case approvable in the first place, and separately eliminated the recurring financial penalties.
Read Hold The Line →Building cost proposals that get approved
Every cost proposal I submitted was approved by leadership. Not because the cases were uncontroversial. Some were. Because each one was structured around three things leadership actually needed: the current cost, the proposed change, and a risk-rated view of what happened if nothing changed. That format meant no follow-up meetings and no requests to come back with more data.
The same leverage, applied to scope
The evidence-first discipline that got these cost cases approved also expanded what the team could cover. Across a fragmented development-partner environment, structured workstreams and an internal triage tool absorbed more work without adding a single hire.
Read Clear The Fog →The results
External Spend Removed
$27k/mo
Contractor dependency eliminated.
Cost Proposals
0 rejected
Each case preempted objections.
Vendors Assessed
15
Reviewed; restructured or exited.
The savings were structural, not one-off cuts. The contractor spend came off because the capability that replaced it was built to last, and the legacy infrastructure came down on evidence rather than argument. Fifteen vendors were assessed, and every cost proposal cleared on the first pass because each one answered the questions before they were asked.
The Big Short edition
What made it hard
Cost cases are political before they are financial. Every line item has an owner, and nobody welcomes the suggestion that their spend is not delivering. The approach has to be evidence-first every time. You present the data and let the room reach the conclusion rather than leading with the recommendation.
The infrastructure decommission required going in person when monitoring data alone was not enough. Some teams claimed they were still using systems with no recorded transactions in months. A direct conversation in the room resolved that faster than any email chain would have.
Removing contractors while maintaining output is a trust exercise. Management needs to believe the team can handle it before they will approve the change. Showing training progress against defined milestones in real time was what made that case land.
Zero rejections was not luck. Each case was built to answer the questions leadership would ask before they asked them. That preparation is invisible in the outcome but it is the entire job.
Know you are overspending but not sure where?
That is exactly where this starts.