Before you build an internal tool, map one job end to end
Most failed internal software was scoped from an org chart instead of a job. One afternoon with a pen fixes that.
Internal software usually fails for the same reason: it was scoped from the org chart instead of from a job. Someone lists the departments, each department lists its screens, and the result is a tool that mirrors how the company is drawn rather than how work actually moves through it. The cheapest fix costs one afternoon and a sheet of paper.
The exercise
Pick one real job that finished recently — not a representative example, an actual one, with a name on it. Then follow it from the first contact to the money landing, and write down every step. For each step, record four things:
- Who touched it. A person, not a role.
- Where it lived. WhatsApp, a sheet, email, a printed sheet on a desk, somebody's head.
- What got typed again. The customer's name, the address, the item list, the price.
- Where it waited. Overnight for an approval, three days for a part, two hours because one person was on site.
Do it with the people who do the work, out loud, in one room. The map is almost never what the owner expects, and the disagreements along the way are the most valuable part of the afternoon.
Reading the map
Three things jump out of every map we have ever drawn:
- Re-entries. Count how many times the same fact was typed. Each one is both wasted minutes and a future discrepancy.
- Waits. Most jobs spend far more time waiting than being worked on. Software rarely makes work faster; it is very good at removing waits, because a wait is usually just somebody not knowing it is their turn.
- Single points of knowledge. Any step that only one person can do, or where the status lives only in their head, is where the business stops when they are away.
What the map usually shows
A twelve-step job where four steps are re-typing, three are waiting for someone to notice, and one person is the only one who knows what has been promised. None of those are solved by a dashboard, which is what everybody asks for first.
Turning a map into a scope
Now the useful part. Build in this order:
- The record first. One place where the job exists, with a status, an owner and a history. Nothing else works without it, and on its own it ends most of the "where are we with this?" traffic.
- The two worst handoffs. Not all of them. The two where things are re-typed or dropped most often. Make the handoff explicit: it becomes someone's turn, and they are told.
- Then, and only then, the reports. Reporting built before the record is reporting on data that nobody is entering yet. It is also the thing most often asked for first, because it is the part the owner personally feels the absence of.
Everything else on the map goes on a second list, dated and parked. Most of it will look less important in three months, and you will have real usage to decide with.
Signs the map was wrong
- It has no waits in it. Every real process has waits; a map without them was drawn from memory of how it should go.
- Everybody agreed immediately. Usually it means the people who actually do the steps were not in the room.
- It describes the best case. Map a job that went wrong as well — exceptions are where the software earns its keep, and where generic products break.
- It starts at "order received". The enquiry that never became an order is part of the job, and it is often where the money is leaking.
Buy, adapt or build
With a map in hand, the decision gets easier. If an off-the-shelf product matches most of the map, buy it and adapt your process at the edges. If the map is ordinary but the details are yours — the way you quote, the stages you use, who approves what — adapting a foundation that already exists is usually the cheapest route to something people will actually open. If the map is genuinely unusual, that is when building from scratch is worth it, and you should expect to pay for it.
Our own products exist for the middle case: Doctor OS for clinics, Navis for maritime crew and port calls, QuickFMS for facilities, PropShare for property enquiries, Hostipal for hospital teams. Starting from one of those shortens the work because the record and the usual stages are already there; the fit to your map is the part we build.
What we would do
The map is the first half of a free 20-minute review, and you keep it whether or not you hire us. Afterwards you get a written scope that lists what we would build first, what we are deliberately leaving for later, and a price — ₹20,000 to ₹1,00,000 for a tool of this size, with where it lands depending on how much of the map you want covered in the first version.
More from the build
- Websites What a business website should do in its first 90 days Traffic is not the first goal. Enquiries are. What to put on the page, what to leave off, and how to tell whether it is working.
- AI search Google is no longer the only place people search Customers now ask ChatGPT and other assistants for recommendations. What makes a site readable to them, and why it overlaps with good SEO.
- Automation The WhatsApp-to-spreadsheet problem Why so many businesses run on chat plus a sheet, what it quietly costs, and the smallest change that ends the retyping.
- AI What an AI pilot should cost, and how to tell if it worked Pick one repeated task, put a number against it before you start, and run it for a month. Everything else is a demo.