How to define MVP scope without building too much
How to build a first release that tests the idea: scope, manual work and essential quality in a job-management service.
Start with the dispatcher’s spreadsheet
A repair company tracks jobs in chat and a spreadsheet. The proposed replacement has grown into three products: a customer portal, a technician app and a dashboard. Before estimating them, sit with a dispatcher and follow yesterday’s jobs. Find the point where a booking was lost, an appointment changed or somebody had to ask for an update.
Choose the problem that makes the current process expensive enough to replace. For this example, it is the handover between dispatchers: the next person cannot reliably tell what is booked and what is finished. The first release should let that person take over a shift without reconstructing it from chat. Customer self-service is a separate question.
Now write the pilot question in terms of behavior: will dispatchers use the service to manage real jobs without keeping their own spreadsheet? Invitations and account creation are easy to count, but they do not answer it. A dispatcher who opens the product every morning and closes jobs there is a more useful signal than fifty people who tried the signup form.
Follow a job all the way to completion
The smallest useful workflow is creation, assignment, rescheduling and completion. Put one actual job through it. The customer calls, the dispatcher books a visit, the technician gets the address and later reports the outcome. Someone on the next shift should be able to see what happened without calling the first dispatcher.
The first version can cover one type of repair, use email and give every dispatcher the same permissions. A technician can report completion by phone while the dispatcher records it. None of these choices requires a separate mobile app. You are reducing the number of things to build while keeping the job moving.
Watch for steps that appear optional on a screen but happen every day. A customer changes the appointment. A dispatcher assigns the wrong technician. An address needs correcting. If ordinary corrections send people back to chat, the workflow is incomplete. Walk through these cases before arguing about the dashboard.
Be honest about work done by hand
Importing old records manually is often a sensible shortcut. So is helping an invited company set up its account. These tasks are outside the daily behavior you want to observe. Tell participants what help is included and keep a record of the time it takes; the future cost of running the service still matters.
Entering every new job on the dispatcher’s behalf is different. The product may contain completed jobs, but you have learned little about whether the dispatcher would choose it. For every useful result, ask who did the work. Independent use and work performed by the founder need separate counts.
Keep the record simple: date, company, task and minutes of assistance. At the end of a week, look at the recurring tasks. Ten minutes of setup may disappear with a better import. An hour spent fixing each job might expose a broken workflow. Automate the work you understand, once its cost or frequency justifies the build.
Do not ask users to risk their records
People will keep a backup spreadsheet if the service loses jobs. That makes the pilot hard to interpret: the team may like the idea and still refuse to depend on it. Protect the records and promises that users rely on from the first release, even when secondary features are rough.
Test the ordinary failures. A double click must not create two jobs. Changing a record URL must not reveal another company’s work. An accidental deletion needs an agreed recovery path. A backup is only useful if you can restore it and find the expected jobs. These are checks on this workflow, not a shopping list of infrastructure.
For a small invited pilot, an operator can handle account recovery or a support request. Set expectations before the first job is entered: who to contact, what happens when the service is unavailable and where the records will go if the pilot stops. Do not leave a participating business to discover those terms during a failure.
Write down what the first release excludes
A useful scope brief fits on a page. It names the dispatchers, the workflow, the supported conditions and the features left for later. Write behavior precisely: “email the technician after assignment” tells the team more than “notifications”. Read receipts, SMS and notification preferences are separate work.
When a new request arrives, check it against an actual job. Correcting an assignment belongs in the first release because dispatchers make mistakes. Automatic matching can wait if a person already knows whom to assign. If something essential was missed, make a trade: remove another task, move the date or change the budget. Record the decision.
Judge readiness against the process being replaced. Shape Up calls this a comparison with the current baseline: here, chat and the spreadsheet. The release can be better at shift handover without matching an ideal scheduling platform. Give the pilot a start date and stop adding features when the supported workflow is ready.
Check what people do without reminders
Run the pilot across a normal working period, including shift changes. Review each company separately: real jobs entered, jobs closed, dispatchers who returned and help required. An average can hide one regular user and several accounts that stopped after setup. Ask people to show their latest job, rather than to rate the concept.
Look closely at work still done elsewhere. Was a needed status missing? Did another employee control the process? Did the service work, but the company have too few jobs that week? Usage counts help locate the question; a conversation and a real example help answer it. Fix the missing step before deciding the audience needs another application.
A free pilot also leaves the commercial question open. Discuss a paid continuation with businesses that return independently, including the price and support they expect. If participation only continues after reminders or hands-on help, find out why. The next release should follow that evidence, rather than revive everything you deferred.
Keep reading
From a business problem to an implementation
From a request for a customer portal to workable scope, a real integration and a measurable business result.
Scaling services: when more instances do not help
Why a service can be slow with idle CPU, how extra instances overload the database and what to check before autoscaling.