You will be able to estimate the monthly cost and payback of an automation, including your own time.
The full monthly cost of an automation is the tool plan plus usage charges, AI charges, any other subscriptions the flow needs, and the value of your time spent keeping it running. Set that total against the time and error cost the flow saves, and you can see whether it pays for itself and how long the build takes to repay.
Most people quote only the tool's monthly plan as the cost of their automation, because that is the charge they see on the credit card statement. The plan price leaves out AI charges, which come on a different bill. It also leaves out the hours you spent building the flow and the time you spend each month checking and fixing it. If you leave those costs out, almost any automation looks like a bargain.
Total each of these per month for the flow you are costing.
Tool plan: the automation tool's subscription. If several flows share one plan, count only this flow's share. If you self-host n8n, the plan cost is replaced by the cost of the rented server plus the time you spend keeping it updated. Usage charges: each tool counts usage in its own unit, as lesson 3.1, "Zapier, Make and n8n: three ways to build the same flow", explained. If your plan includes an allowance, this flow uses part of it. If you go over, you pay extra or move to a bigger plan. AI charges: every AI step costs money per run. You pay either through the tool's own usage or through the model provider's bill if you use your own API key. The amount depends on how much text each run sends and receives, so check the provider's current pricing page and your usage dashboard instead of guessing. Other subscriptions: any service that exists only for this flow, such as a form tool, a document generation service or a paid messaging app.
Your own time counts as well, both the time to build the flow and the time to keep it running.
Build time is a one-off cost. It covers the hours you spent designing, building and testing, from your module 2 diagram through to the failure tests in lesson 7.4, "Break your own flow on purpose". Add those hours up.
Maintenance is a monthly cost, and it is never zero. Connections expire, someone renames a folder, the AI step's prompt needs adjusting, and the review queue needs clearing. Make a realistic monthly allowance for checking the weekly summary from lesson 7.2 and fixing what it shows. One hour a month is a modest starting guess for a small flow. After a few months, the log from lesson 7.3, "Logs and human review checkpoints", will show the real maintenance figure, and you can correct your allowance from it.
To turn hours into dollars, you need an hourly value for your time. Use your salary divided by your working hours, or what you would pay someone else to do the task. Pick one figure and use it consistently on both the cost side and the savings side.
Your savings figures come from the scoring sheet in lesson 1.4, "Score your own task list", which records the monthly minutes the task took by hand and the estimated error cost. Time saved is the manual time minus the time the flow still needs from a person, such as reviewing drafts or clearing the review list. Convert the time saved to dollars at your hourly value.
Subtract the monthly costs, including maintenance, from the monthly value of time saved to get the net monthly saving. Then divide the build cost by the net monthly saving to get the payback period in months.
Mei Ling built a flow for the enquiries she handles at the centre. All the figures here are invented examples, not real prices or rates. The centre gets about 15 enquiries a week, about 65 a month. Lesson 1.2's process fix had already brought each enquiry down to about 3 minutes by hand. With the flow, she spends about half a minute per enquiry checking the draft, so she saves 2.5 minutes on each. Across 65 enquiries that comes to 162.5 minutes, or about 2.7 hours a month. She values her time at S$25 an hour, so the time saved is worth about S$67.70 a month.
Her costs are a S$30 a month plan, AI charges of S$0.02 per run, which is S$1.30 for 65 runs, and 1 hour of maintenance a month, worth S$25. Her total monthly cost is S$56.30 (S$30 + S$1.30 + S$25), leaving a net saving in time of about S$11.40 a month (S$67.70 - S$56.30).
The build took about 8 hours, worth S$200 at S$25 an hour. On time savings alone, S$200 / S$11.40 means about 17 or 18 months before the build is repaid, which is a weak case.
Time was never her main reason for choosing this task. In lesson 1.1 she identified the error cost: slow replies in busy months were losing students. Preventing even a few lost enrolments a year would outweigh the time figures. A weak case on time alone can still be worth it if the flow prevents a costly error, so write the error cost next to your numbers, in dollars or in plain words, and base the decision on both time and error cost.
With usage-based charging, whether by the tool or the model provider, cost grows with volume. Benefits usually grow with volume too, but not always at the same rate. If Mei Ling's enquiries doubled to 130 a month, the time saved would be worth about S$135.40 and her costs would be about S$57.60, so the flow would pay much better, assuming the plan's allowance still covered the extra runs. A plan's allowance can run out suddenly, and if the extra volume forced her onto a larger plan, the result could swing the other way. Rerun the sums whenever volume doubles or a tool changes its pricing.
In the activity below you will fill in a cost sheet for your main flow with the plan, usage, AI and maintenance costs, and calculate the monthly payback. Use the real figures from your accounts and pricing pages where you have them, and label every estimate as an estimate.
Fill in a cost sheet for your main flow with plan, usage, AI and maintenance costs, and calculate the monthly payback.
Junxiong-WFG Organisation is an authorised representative of AIA Financial Advisers Private Limited (Reg. No. 201715016G).