Most operations teams can list ten processes that should be digitized: approvals by email, orders copied between systems, reports assembled by hand, onboarding checklists in a shared folder. The list is rarely the problem. The problem is the meeting where everybody argues for the process that annoys them most, and the first project is chosen by whoever speaks last.
This article is for a manager who has to pick the first process to digitize and wants a reason that colleagues will accept. It describes a one-page scoring table with three measures: hours, errors and owner. It is deliberately simple. If your candidates involve AI specifically, our article on prioritizing AI use cases for business teams covers that narrower case. Here we talk about processes in general.
Why score at all
Without a shared method, the decision depends on who is loudest or who has the most senior title. A table does not make the decision for you, but it moves the discussion from opinions to numbers that people can check. It also leaves a record. Six months later, when someone asks why the invoice process came before the onboarding process, the page answers.
The three measures
Hours. How many person-hours does the process take each week, counting everyone involved? Count the real work, including the time spent chasing people and correcting things. Estimates are fine, but ask the people who do the work, not the people who manage it. Our article on the true cost of manual processes shows how hidden time adds up.
Errors. How often does something go wrong, and what does it cost when it does? Count mistakes you can recall from the last three months: a wrong delivery date, a duplicate payment, a missing approval. Give weight to errors that reach a customer or a regulator, because they cost more than errors caught internally.
Owner. Is there one named person who understands the process, cares about it and has the authority to change it? This measure is not about the size of the gain, but it is the best predictor of whether the project ends. A process with no owner will stall when the first disagreement appears, however large the saving looks.
How to build the table
Create a table with one row per candidate and these columns: process name, weekly hours, errors in the last three months, owner, and a score for each measure from 1 to 3.
- Hours: 1 for under two hours a week, 2 for two to ten, 3 for more than ten. Adjust the bands to the size of your team, but keep them fixed for all candidates.
- Errors: 1 for no remembered errors, 2 for a few minor ones, 3 for repeated errors or any that reached a customer.
- Owner: 1 for nobody, 2 for someone willing but without authority, 3 for a named person who can decide.
Add the three scores. The maximum is 9. Do not weigh the measures at first. If you want to give owner double weight, do that after the first pass and see whether the order changes.
A short example
The following example is invented for illustration and describes no real company. A team of twelve lists four candidates.
| Process | Hours per week | Errors | Owner | Total |
|---|---|---|---|---|
| Copying orders from email into the order system | 12 hours (3) | Several wrong dates (3) | Head of customer service (3) | 9 |
| Monthly report assembled from three spreadsheets | 6 hours (2) | A few minor mistakes (2) | Finance lead, little time (2) | 6 |
| Employee onboarding checklist | 3 hours (2) | None remembered (1) | HR, willing but no authority over IT (2) | 5 |
| Equipment requests by phone | 1 hour (1) | None remembered (1) | Nobody (1) | 3 |
The order copying process wins clearly. Notice that the monthly report would have been the favorite of the finance lead, and it would probably have been a fine project. The table does not say it is a bad idea. It says that something else is bigger, riskier to leave alone and has a person ready to lead it.
How to fill it in within an hour
Ask three people who do the work to join you for sixty minutes. Spend the first twenty on the list of candidates, with no scoring. Spend the next twenty on hours and errors, one process at a time, writing the number everyone can live with. Spend the last twenty on owners: for each process, say a name or write "nobody". Naming nobody is useful information.
Do not try to be exact. If two people disagree on hours by a factor of two, write the lower number and note the doubt. A rough table that exists is better than a precise one that is still being prepared next month.
When the scores are close
If two processes end with the same total, choose the one with the clearer owner. If owners are equal, choose the one whose result is easiest to check afterwards, such as a number of hours saved or errors avoided. Do not try to do both at once. Two simultaneous projects in a small team usually mean two delayed projects. For a view of how a first project fits into a longer plan, see automation in 90 days.
Common mistakes
- Scoring by enthusiasm. Count hours and errors that someone can point to, not how much a process irritates people.
- Ignoring the owner. A big saving without an owner often ends as an unfinished project.
- Choosing the most complex process first. If two candidates are close, the simpler one gives a visible result sooner.
- Hiding the table. Share it with everyone who took part, so the decision looks fair.
- Treating the score as final. Review it after each project, because hours and owners change.
What to do this week
List your candidates, score them in one hour with the people who do the work, and pick one. Write one sentence for each process that was not chosen, saying why it waits. Then name the owner of the winner and agree what result you will check in a month. If the process involves an old system, modernizing legacy systems may help you plan the next step.
Sources
- Our related articles linked above.
- The scoring bands and the example are a general working approach and are invented for illustration; they do not describe a particular company or system.
