Application portfolio
Mangrove runs its partner business on about ten applications, and nobody has ever seen them on one page. This lesson shows you how to list them, lay them over the capability map to find duplicates and gaps, score each one's business and technical fit with evidence, and turn the scores into keep, invest, replace or retire decisions that say when, why and at what risk.
By the end you can
- Build an application inventory with owners, capabilities, users, yearly cost, support dates and the data each system owns
- Map applications onto the capability map and name the duplicates, gaps and misfits
- Score business fit and technical fit from evidence and place each application in the TIME model
- Work out what an application really costs a year, including the costs its users carry
- Write portfolio decisions that account for timing, dependencies and risk
The everyday picture
A family is about to move into a smaller flat, and the kitchen has to be sorted first. They empty every cupboard onto the table and group things by job: boiling water, cutting, storing, cooking. Two kettles sit side by side. There is no proper bread knife, so everyone saws bread with the kitchen scissors. The fridge keeps food cold, but its maker stopped selling spare parts last year.
For each item they ask two questions. Does it do its job well? Is it in good condition and still repairable? The good kettle stays, the leaky one goes. The fridge works, so it is replaced on purpose before it breaks in the middle of summer. A good knife is bought for the gap. And nobody throws out the old kettle until the new one has boiled water. An application portfolio review is the same job for a company's software: everything on one table, grouped by job, two questions for each item, and a decision with a sensible order.
The application inventory
The capability map from Capability maps says what Mangrove does. The value stream from Value streams and processes says how partner orders flow from order to cash. Neither says which software does the work. That is the job of an application inventory. TOGAF calls a list like this an application portfolio catalogue and uses it in the information systems architecture work of Phase C.
Record the same facts for every entry, so entries can be compared:
- Business owner and technical owner. The business owner decides what the system should do and answers for it; the technical owner keeps it running. An entry without a business owner is the first warning sign.
- Capabilities supported, using the names on the capability map, never team or project names.
- Users: who, and roughly how many. "About 40 sales reps" is useful; "sales" is not.
- Cost per year, all-in: licences or subscription, hosting, support contracts and the people who run and change it.
- Technology and support end date: the stack or product version, and when the vendor or the last skilled person stops supporting it.
- Data it owns: which records it is the official source for, and which it only copies.
Find entries in four places. Finance's list of software contracts and invoices finds everything that is paid for. Cloud and hosting accounts find what is running. The single sign-on list finds what people log in to. Then ask each team which screens and files they open on a normal day. Only the last question finds the spreadsheets and chat groups, and at Mangrove some of the most important work lives there: credit approvals run on email and a spreadsheet. Count a spreadsheet as an application when people rely on it to do a job of the business.
Keep the scope small. Start with the area the goals point at, here the partner business with about ten entries, instead of listing all of Mangrove's systems first. And give the inventory an owner and update dates: contract renewals, support end dates and major changes. An inventory that nobody maintains is wrong within a year.
Quick check
You build the inventory only from finance's software contracts and the hosting accounts. Which entry are you most likely to miss?
The spreadsheet has no contract and no server of its own, so only asking people how they work finds it. The agency, the WMS and the ERP all send invoices, so a contract list catches them.