Workflow Automation With Custom Software: Replacing Spreadsheets, Email and Manual Handoffs

Prices are in US dollars. Euro and pound figures in parentheses are approximate, rounded conversions. Quotes are issued in the currency of your contract.

Most companies do not have a workflow problem. They have a handoff problem. The ERP works, the CRM works, the warehouse system works, and between them sit people forwarding emails, copying numbers from one screen to another and maintaining spreadsheets that nobody designed. The work gets done, slowly, and the errors appear downstream where they are expensive. This post is about finding those handoffs, deciding which ones to automate first, and building automation that the team trusts rather than works around.

Where do manual handoffs hide?

• In email. Approvals, requests and exceptions travel as messages with attachments. Nobody can see the queue, and a request waits as long as the inbox it landed in.
• In spreadsheets. A tracker that someone updates from three systems every morning. It is the real status board for the process, and it is wrong by lunchtime.
• Between systems. A sales order confirmed in the CRM is typed into the ERP. A delivery confirmed in the warehouse is typed into the invoicing tool. Each re-keying is a delay and a chance to make a mistake.
• In people's heads. The rule that says which customers get credit, which orders need a second check, and which supplier to use when the first is out of stock. It lives with one person and leaves with them.
• At month end. Reporting that takes a week because the data has to be assembled from exports. The report is a symptom: the process that produces the data is not connected.

The way to find them is not a workshop. It is sitting beside the people doing the work for a day and writing down every time they switch screens, open a spreadsheet or send a message to move the work along. Most companies find fifteen to thirty such steps in a single core process.

Which workflows pay back fastest?

Payback is measured in hours saved and errors avoided. A workflow that takes ten people two hours a day is roughly a full-time salary a year before the first error is counted. Workflows with few people and rare errors are not worth automating yet, however annoying they are.

How do you map a process before automating it?

1. Record the real path. Follow five real orders, requests or cases from start to finish, with timestamps. Note every system touched, every person involved and every wait.
2. Separate the rule from the exception. Most steps follow a rule that can be written down. The exceptions are where the expertise is, and they are the part the software must hand to a person.
3. Find the system of record for each piece of data. The customer lives in the CRM, the order in the ERP, the stock level in the warehouse system. Automation reads from and writes to those, and never keeps its own copy as the truth. What is a database is a useful primer for the non-technical sponsor.
4. Measure the baseline. Time per case, cases per week, errors per month, cost per error. Without this you cannot show the result.
5. Decide what not to automate. Steps that happen twice a year, decisions that need judgement, and anything the team cannot explain. Those stay manual for now.

The output is a one-page map of the workflow with the manual handoffs marked, the systems named and the exceptions listed. That document is the scope. It is also most of what a discovery phase produces for a larger program.

What does the software actually do?

A workflow automation tool has four parts. It listens for events in your systems, a new order, a signed contract, a delivery scan. It applies rules to decide what happens next. It writes the result into the right system, through its API rather than through a person's screen. And it shows people a queue of the cases that need them, with everything they need to decide on one page. What is an API explains the plumbing that makes the first and third parts possible.

The temptation is to build a grand platform. The right first build is one workflow, end to end, in three to five months, for $80,000 to $200,000 (€70,000 to €175,000, £60,000 to £150,000). That is enough to prove the approach, pay back, and teach the team what the second workflow should be. Enterprise application integration covers the integration layer that the second and third workflows can share.

How do you keep people in the loop?

• Automate the routine, queue the exception. The software handles the 90 percent that follows rules and puts the rest in front of a named person with context.
• Make every automated action visible. A log of what the system did, why, and on whose rule. When something goes wrong, the team can find it in minutes.
• Make every action reversible. A person can hold, redo or override any step. Automation that cannot be stopped is automation the team will switch off.
• Start in shadow mode. For the first weeks the software proposes and a person confirms. When the proposals are right, the person steps back.
• Let the team own the rules. Thresholds, routing and escalation times should be editable by the operations lead, not by a developer.

The people who ran the manual process are the ones who know where it breaks. Involve them in the design and the tool will be adopted. Impose it and the spreadsheet will quietly reappear. User adoption for internal software goes deeper on the rollout.

How do you measure the result?

How 7L approaches this

7L International designs, builds and supports custom software from business hubs in New York, London and Athens. We were founded during the Greek financial crisis, when doing more with the same people was the only option our first clients had, and workflow automation has been part of our work for more than one hundred organizations since, from hospitality operations to fleets and smart buildings. We start with a day beside your team, not with a platform pitch, and we build one workflow at a time on top of the systems you already run. The enterprise software development guide covers how that fits into a larger program. If you have a process held together by email and a spreadsheet, describe it to us and we will tell you whether it is worth automating yet.