An enterprise has an application portfolio of several hundred systems and cannot decide what to retire. What analysis makes the decision tractable?
Show the full answer Hide the answer
The analysis
Two dimensions, assessed per application, producing four quadrants:
Business value — how important is what this application does, measured by the capability it supports and the criticality of that capability.
Technical condition — supportability, technology currency, integration burden, incident rate, cost to operate, and key-person dependency.
| Low technical condition | High technical condition | |
|---|---|---|
| High business value | Invest or replace — the most urgent quadrant | Maintain |
| Low business value | Retire | Tolerate or consolidate |
The evidence that makes it tractable
The decision stalls because nobody can defend a retirement. That is an information problem with a technical solution:
- Actual usage data. Logins, transactions, API calls over ninety days. A substantial number of enterprise applications turn out to have almost no use, and measured evidence is what makes retirement defensible rather than courageous.
- Dependency mapping, ideally derived from traffic and query logs rather than from self-declaration. Most fear of retiring an application is fear of unknown consumers.
- Cost per application, including licences, infrastructure, and the engineering and support time consumed.
- The capability it supports, so overlap with other applications is visible.
The sequencing that works
1. Retire the unused first. Low risk, immediate saving, and it builds credibility for the harder decisions.
2. Consolidate duplicates within a capability, where the capability map has already shown overlap.
3. Address high-value, poor-condition systems with a business case built on the cost of inaction — incident cost, change lead time, features refused — rather than on technical virtue.
4. Leave the tolerable alone. Not everything needs to be modernised, and an application that works, costs little and changes rarely is not a problem.
The failure mode
Portfolio analysis as an annual exercise producing a report. Without an owner, a decision forum and a funded decommissioning capability, the analysis is repeated each year with similar findings.
Decommissioning must be a funded deliverable with a date and an owner, because it is unglamorous work that loses to every alternative unless it is protected — and an estate that only ever adds applications is the default outcome.