What to Automate First: A Simple Scoring Method for Busy Managers
Everyone has a list of tasks they want automated. A five-factor scoring method helps you pick the ones that will actually pay back, and spot the ones to leave alone.
Once a management team decides to invest in automation, the requests arrive fast. Finance wants bank reconciliation automated, sales wants quotations generated from the CRM (customer relationship management) system, HR (human resources) wants leave requests off paper, and the owner wants a daily dashboard on his phone. All of them sound reasonable, and you cannot do them all at once.
Picking the loudest request, or the one the vendor demonstrated best, is how automation budgets get spent on things that save little. A simple scoring method for deciding what to automate first gives you a fairer way to choose, and the scoring itself takes an afternoon rather than a study.
The five factors that matter most
Score every candidate process from 1 (low) to 5 (high) on each of these factors. Keep the definitions consistent so the comparison is fair.
| Factor | Question to ask | Scores high when |
|---|---|---|
| Volume | How often does this happen? | It runs many times a day or week, not once a quarter |
| Manual effort | How much human time does each run take? | Each run involves typing, copying, checking or chasing |
| Error cost | What happens when it goes wrong? | Mistakes reach customers, suppliers, payments or compliance |
| Stability | Is the process clear and consistent? | Everyone does it the same way and the rules are written down |
| Data readiness | Is the input data digital and reliable? | Inputs come from a system or structured form, not handwriting or free text |
Add the five scores for a total out of 25. Then apply one adjustment: if stability or data readiness scores 1 or 2, put the process on a separate “fix first” list, whatever its total. Automating an unstable process or one fed by messy data creates more work, not less.
A worked example from a distribution company
Consider a medical supplies distributor in Amman serving pharmacies and private hospitals. The operations manager has four candidates on the table and scores them with the finance and sales leads in one meeting.
| Process | Volume | Effort | Error cost | Stability | Data | Total |
|---|---|---|---|---|---|---|
| Purchase order approvals | 4 | 3 | 3 | 4 | 4 | 18 |
| Daily sales summary to owner | 5 | 3 | 2 | 5 | 4 | 19 |
| Customer credit checks before dispatch | 5 | 4 | 5 | 2 | 3 | 19 |
| Supplier invoice matching | 4 | 5 | 4 | 4 | 2 | 19 |
Three processes tie on 19, which shows why the totals should not decide alone. Credit checks score high on value but low on stability, because each sales rep applies the credit policy differently. That makes it a process redesign job first. Invoice matching scores 2 on data, because many supplier invoices arrive as scanned PDFs with inconsistent layouts, so it also goes on the fix-first list until the inputs improve.
That leaves the daily sales summary and purchase order approvals as the first two projects. Both are stable, both use data that already sits in the ERP (enterprise resource planning) system, and both save time every single day. Neither is glamorous, but both are likely to work, which builds trust for the harder projects later.
How to run the scoring session
- Collect candidates from the people who do the work. Ask each team for the three tasks that waste most of their week. You will get a better list than one written by managers alone.
- Describe each candidate in one sentence. For example, “Approving purchase orders above 500 dinars by email between the requester, department head and finance”. Vague candidates such as “finance automation” cannot be scored.
- Gather rough facts before the meeting. How many times a month does it run? Roughly how long does each run take? What went wrong in the last quarter? Estimates are fine, but they should come from the people involved.
- Score together, out loud. Put the table on a screen with representatives from operations, finance and IT (information technology). Disagreements over a score usually reveal something important, such as two departments doing the same process differently.
- Separate the results into three lists. Automate now, fix first, and leave alone. The last list matters too. Some processes are too rare or too variable to be worth automating.
- Revisit every quarter. Once a fix-first process is redesigned and its data cleaned, score it again. It may move straight to the top.
Turning a score into a business case
A score ranks the options, but a budget holder will still ask what the project returns. For the top candidates, build a simple estimate. Take the number of runs per month, multiply by the minutes saved per run, and convert to hours. Add the cost of errors that automation would realistically prevent. Compare that with the build cost and any monthly licence fees.
Be honest about what does not go away. An automated approval still needs someone to approve. An automated report still needs someone to read it and act. Count the time saved on chasing, copying and checking, not the whole process. A conservative estimate that turns out right is worth far more to your credibility than an optimistic one that does not.
Also allow for upkeep. Every automation needs an owner who notices when it fails, updates it when a form or system changes, and answers questions from users. Small automations nobody owns tend to break quietly within a year.
Share the scored table with the teams who suggested candidates. People accept that their request is not first when they can see why, and they often come back with better information that changes a score. The table then becomes a running list that the teams can see, rather than a one-off decision made behind closed doors.
Choosing well beats choosing fast
You can run the first scoring session within a fortnight: one week to collect candidates and rough volumes from each team, then a two-hour meeting to score them. Expect the fix-first list to be as useful as the automate-now list, because it tells you which processes need redesign before any tool touches them.
This is how we approach the Automate stage of the KEMLAT Method: automate what truly deserves it, once the process is understood. Our Intelligent Automation service helps you build the candidate list, score it honestly, and deliver the first automations in a way your team will trust and maintain.