You inherit an estate of roughly 400 applications, no reliable inventory, and a mandate to reduce cost and risk. What do you do in the first ninety days?
Show the full answer Hide the answer
What the interviewer is testing
Whether you can sequence work at portfolio scale, and whether you go for evidence before strategy. This is the enterprise architect's opening problem and it has a well-worn right answer.
Days 1–30: get an inventory
You cannot decide anything about 400 applications you cannot list. Build the minimum viable record per application — and resist the temptation to gather forty attributes, because that survey never completes:
| Field | Why |
|---|---|
| Owner (a named person) | Every later decision needs someone to make it |
| Business capability supported | The stable frame for grouping and comparison |
| Annual run cost | Infrastructure plus licence plus support |
| Active users | Separates "critical" from "somebody's spreadsheet replacement" |
| Technical health | Supported versions, security posture, dependency currency |
| Business criticality | Rated by the business, not by IT |
Source it from what already exists — the CMDB, cloud tags, licence agreements, SSO login data, network flow logs — before asking people. Login data in particular is the fastest honest answer to "is anyone actually using this".
Days 30–60: find the cheap wins
Two analyses, both of which produce action rather than a report:
Retire. Applications with near-zero usage, superseded systems still running, environments that outlived their project. This is consistently the largest immediate win and the one nobody has time to do — it removes run cost, licence cost, security exposure and support burden at once, with no migration.
Duplication. Overlay applications onto the capability map. Five systems doing one capability is visible instantly on a map and invisible on an org chart. Consolidation is slower than retirement, but it is where the structural cost sits.
Days 60–90: build the frame for everything after
Value against health grid. High value, poor health → invest or replace. Low value, poor health → retire. High value, good health → leave alone. Low value, good health → leave alone and stop funding enhancements.
Assign a disposition to every application — retain, replatform, replace, retire — with an owner and a date, even where the answer is "review in 12 months".
Fix the inflow. Rationalising a portfolio while new unmanaged applications keep arriving is bailing without plugging the hole. That means a lightweight intake with a real threshold, and an owner recorded before anything reaches production.
What to be honest about
Ninety days does not reduce cost much; it produces the inventory, the retirement list and the prioritised plan. The retirements start paying back in months four to twelve. Promising savings in the first quarter is how these programmes lose credibility in the second.
What a strong answer adds
Noting that the inventory decays immediately unless it is fed by automation — cloud tagging enforced at provisioning, SSO application registration, pipeline registration — because a hand-maintained portfolio record is accurate exactly once.