Measure a Process Before Automating It
Learn how to measure a process over two weeks using handling time, task volume, and rework counts, so you can identify bottlenecks and decide which steps to automate first.
Automation often begins with a familiar feeling: this task repeats too often, staff lose too much time on it, and software should make it faster. But that feeling does not yet show whether to automate the entire process or just one step, nor what the result should change. Before choosing a tool, a business should measure how the process currently runs. Two weeks of focused note-taking is enough to support a decision, instead of debating based on each person’s impression.
Measure to find where the problem sits
Seen from the outside, a process may involve many tasks: receiving a request, checking information, entering data, sending it for approval, notifying the result, and filing records. The feeling of “taking too much time” usually merges all of these, while the real cause may sit in re-entering data or waiting for someone else to supply information. If the whole chain is automated at once, the business can easily complicate steps that were never a problem.
Measurement should keep the current way of working unchanged; do not measure and redesign at the same time. The person responsible should choose one specific workflow, such as handling purchase orders or receiving internal requests, and record it from the start until the file is complete. The measurement is not meant to judge individual productivity; it is used to find repeated steps, waiting points, and points where errors are likely.
The goal is not to prove that the process is slow, but to identify which step is slow and why.
Three numbers to record
No time-tracking system or complex form is needed. A shared spreadsheet is enough, as long as each run is recorded as one row and everyone reads the columns the same way.
- Time per run: Record the time the user actually spends working, from opening the file to moving it to the next step. Keep waiting time for a response or approval separate, because a long wait does not mean the person handling it works for a long time. If a run is interrupted, add a note instead of estimating a single overall figure.
- Number of runs: Count the files or requests that pass through that step in two weeks. An action that takes little time but occurs constantly can still deserve attention before a rare task. Include runs done outside working hours or handled by someone else, so actual volume is not missed.
- Number of corrections: Count each time a file has to be reopened to fix data, add attachments, change the recipient, or be reprocessed after being moved on. Write the cause briefly next to each correction. Errors from missing input information and errors from manual entry are two different problems, and they need different automation approaches.
The three numbers should be read together. Time per run multiplied by the number of runs shows the volume of repeated work; the number of corrections shows the cost of doing work wrong and redoing it. Do not simply pick the step that consumes the most hours if that step depends on professional judgment while the errors actually arise in a simple data-entry action earlier in the chain.
How to measure over two weeks without disrupting the team
Before starting, agree on the start and end points of the process. For example, “handling a purchase request” might begin when a complete form is received and end when the request is passed to the approver, rather than including the whole delivery stage. A clear scope makes the recorded rows comparable with each other.
During the two weeks, the person doing the work only needs to update the sheet right after each run: the time of the work, the time spent on actions, any waiting time, whether a correction was needed, and the cause. At the end of each day, one coordinator checks rows with missing information but does not fill them in with guesses. Special cases such as urgent files, missing documents, or changed requests should be flagged separately.
Two weeks is enough for an initial decision on which step to prioritize; it is not meant to prove that every process runs the same way all year. If the measurement period overlaps a holiday, a staffing change, or seasonal work, record the context and consider measuring again before a large investment. At the end, the team only needs to answer: which step repeats most, which step creates the most corrections, and is the data at that step stable yet.
Choose the one step most worth automating
A suitable step usually has clear inputs, stable processing rules, and results that are easy to check. For example, software can generate a file number, fill known information into a form, check required fields, or send a notification when conditions are met. Conversely, a step that requires professional judgment, negotiation, or an exception decision should not yet be forced into a rigid flow.
- Prioritize steps that occur frequently and take up a large share of working time.
- Prioritize copying, matching, or re-entering structured data.
- Temporarily exclude rare cases so that the owner handles them manually.
- Keep one checkpoint before data is recorded or sent.
After choosing the step, rewrite the intended flow in operational language: where the data comes from, which conditions are processed, which cases must stop, who receives notifications, and where users check the result. When the solution is implemented in management software, compare the trial results with the figures already recorded. If the action is faster but errors are harder to detect, or staff still have to re-enter data somewhere else, the design has not solved the right problem.
What to do now
- Choose one specific process and build a sheet with three columns: time per run, number of runs, number of corrections.
- Keep recording for two weeks, flagging waiting time and exceptions separately.
- Pick a single repeated step with clear rules and easy checks to test automation first.
Measuring first lets the automation discussion start from operational facts rather than a feature list. When a process needs to be reviewed and a suitable pilot scope defined, VASCOM can work with the business starting from that same record sheet.