- automation roi
- planning
- process improvement
Measure Before You Automate: The One-Week Baseline That Makes ROI Provable
Most automation ROI claims are guesses because nobody measured the manual process first. Here is a simple one-week baseline, the four numbers to capture, and a worked example with real math.
Kamal Farooqi5 min read
The ROI Number You Can't Prove Later
Someone says an automation will save 20 hours a week. Three months after launch, you ask: did it? Nobody knows. The old process wasn't measured, so there is nothing to compare against. The project feels like it worked, or it doesn't, and either opinion is just a feeling.
This is the most common gap we see in small business automation. It's fixable in about a week, before you build anything, and it takes little more than a notepad or a spreadsheet.
The Four Numbers Worth Capturing
You don't need a time-and-motion study. You need four numbers for the process you plan to automate:
- Volume. How many times does this task happen per month? Pull it from your inbox, CRM, or accounting system rather than guessing.
- Time per task. How many minutes does one instance take, start to finish, including the interruptions?
- Error and rework rate. What percentage need to be fixed, redone, or chased? And how long does a fix take?
- Delay. How long does the item sit before anyone touches it? This is often the number that matters most to customers, and it never shows up in a labor calculation.
Multiply the first three by your loaded hourly cost (wages plus taxes and benefits, usually 1.25 to 1.4 times the base rate) and you have a monthly cost for the process as it stands today.
How to Collect Them in a Week
- Pull volume from systems, not memory. Count invoices sent, forms submitted, or tickets opened over the last three months and take the average.
- Time 10 real instances. Have the person doing the work note start and stop times on ten tasks. Ten is enough to be useful. Don't use their estimate; people usually guess low because they forget the small interruptions.
- Keep a simple error log. For one week, every time something has to be fixed, write down what it was and how long it took.
- Record the delay. Compare the timestamp when the work arrived with the timestamp when it was completed. Most tools already hold both.
Total effort: maybe two to three hours spread across the week. That's cheap insurance on a project that may cost several thousand dollars.
A Worked Example
Take a composite, invented scenario: a small distributor that enters supplier invoices into its accounting system by hand. The baseline week gives us:
- 400 invoices a month
- 6 minutes per invoice to key in
- 8% need rework (wrong totals, wrong vendor), at 15 minutes each
- Loaded labor cost of $32 an hour
The automation reads the invoice, fills in the fields, and flags anything unusual for a person to review. After a month in production, the same measurements show 1.5 minutes per invoice (review only) and a 2% rework rate.
| Measure | Before | After |
|---|---|---|
| Entry time (400 invoices) | 40 hrs | 10 hrs |
| Rework (32 vs 8 invoices at 15 min) | 8 hrs | 2 hrs |
| Total monthly hours | 48 hrs | 12 hrs |
| Monthly labor cost at $32/hr | $1,536 | $384 |
That's 36 hours and $1,152 saved per month. Subtract $120 a month in software and monitoring costs and the net is $1,032. With a $5,500 build, payback is about 5.3 months. Every number in that calculation came from something observed, not assumed.
Without the baseline, you'd have said the automation "saves a lot of time" and been unable to defend the spend at budget season.
Where Baselines Go Wrong
Measuring only the happy path. If your timing sample is ten easy tasks, you'll understate the real cost. Include the messy ones, because those are often what you're trying to eliminate.
Ignoring the delay number. A quote that sits for three days before it's sent loses deals. That cost doesn't appear in labor hours but may be worth more than the hours themselves. Track it and put a rough value on it separately.
Counting hours saved as cash saved. If the person who saves 36 hours isn't reassigned to something valuable or doesn't reduce overtime, you haven't banked money. Decide up front where those hours go.
Skipping the follow-up. Measure again at 30 days and again at 90. Early numbers are often flattering because everyone is paying attention. The 90-day number tells you whether the gain held. It also catches the drift when exceptions pile up and people quietly return to doing things by hand.
What the Baseline Also Gives You
Beyond proving ROI, a baseline helps you choose between projects. When you've measured two or three candidate processes, the one with the highest monthly cost and the most predictable rules usually wins. The decision stops being about which task is most annoying and becomes a comparison of numbers.
It also tells you when not to automate. If the baseline shows 5 hours a month of work, even a perfect automation won't pay back a meaningful build cost. Better to learn that in week one than month six.
Start With One Process and One Week
Pick the task you suspect is costing you the most, and track those four numbers for a week. If you'd like a second pair of eyes on what you find, or help turning the numbers into a payback estimate, book a free call and we'll walk through it together.