Percentages are where chatbot return on investment claims go to become meaningless. "300% ROI" tells you nothing about your business, your prices or your visitors, and it is usually built on someone else's numbers.
Payback period is plainer. How many months of running the assistant before the money it brought in covers everything it cost, including your time? You can work it out on the back of an envelope, and more importantly you can argue with every figure in it.
What follows is one worked chatbot payback period calculation using hypothetical numbers. Every figure is labelled so you can replace it.
The formula, without decoration
Payback in months = total cost to reach steady state ÷ net monthly gain.
Total cost to reach steady state is setup time plus the first month's content writing, costed at a real hourly value. Net monthly gain is the value of extra work won each month, plus any time saved that you genuinely reuse, minus the ongoing monthly cost (subscription and review time).
If net monthly gain is zero or negative, there is no payback period. That is a valid answer and worth knowing early.
Building it step by step
Take a hypothetical kitchen fitting business. Replace each figure with yours.
- Price your time. Say an hour of the owner's time is worth $45.
- Cost the setup. Five hours teaching and installing, plus six hours writing answers: 11 hours, or $495.
- Cost each month. Starter at $20 a month, plus 25 minutes of review a week (about 1.8 hours a month, or $81). Monthly running cost: $101.
- Estimate extra enquiries. The business gets 60 website visitors a month who show clear buying intent. Say the assistant captures 6 enquiries a month that would otherwise have left without contacting anyone. Be strict here: these are enquiries the contact form was not catching.
- Apply your real conversion rate. If one in four enquiries normally becomes a quote that is accepted, 6 extra enquiries means 1.5 extra jobs a month.
- Use margin, not revenue. Say an average job brings $1,200 of gross margin. 1.5 jobs is $1,800 a month.
- Calculate. Net monthly gain is $1,800 − $101 = $1,699. Payback is $495 ÷ $1,699, or roughly 0.3 months.
That result looks absurdly good, which is the moment to be suspicious rather than delighted.
Almost all of that speed comes from step 4. Six genuinely extra enquiries a month, from a site with 60 serious visitors, is an ambitious assumption. It might be true for a business whose contact form is buried at the bottom of a long page, or whose enquiries mostly arrive in the evening. It is unlikely for a business that already replies within the hour and has a prominent phone number. The formula is fine. The input is doing all the work.
Three scenarios, not one
A single estimate hides the part that matters: how wrong could step 4 be? Run the same business through a pessimistic and a middle case.
| Pessimistic | Middle | Optimistic | |
|---|---|---|---|
| Extra enquiries a month | 1 | 3 | 6 |
| Conversion to accepted job | 1 in 6 | 1 in 5 | 1 in 4 |
| Extra jobs a month | 0.17 | 0.6 | 1.5 |
| Margin per job | $900 | $1,100 | $1,200 |
| Monthly margin gained | $150 | $660 | $1,800 |
| Monthly running cost | $101 | $101 | $101 |
| Net monthly gain | $49 | $559 | $1,699 |
| Payback on $495 setup | about 10 months | under 1 month | under 1 month |
The pessimistic column is the one to stare at. One extra enquiry a month, converting badly, still pays back inside a year. If that is true for your figures, the decision is low risk. If your pessimistic column shows a negative net gain, you have learned something important for free.
For a high-value trade the maths is forgiving. For a business selling $60 appointments with thin margins, the same table can look very different, and review time becomes the deciding cost.
Try it quickly. Say a beauty salon makes $25 of margin on an average appointment and gains 4 extra bookings a month in the pessimistic case. That is $100 of margin against $101 of running cost: a net gain of minus $1, and no payback at all. The same salon only gets there by reducing review time, winning repeat visits from those new clients, or both. Repeat business is a legitimate input, but put it in as its own row so you can see how much of the result depends on it.
The assumptions that break the model
A payback model is only as honest as its weakest input. These are the ones that tend to flatter the result.
- Counting every captured lead as extra. Many visitors who talk to the assistant would have filled in the form or rung anyway. Only the leftover counts. Leads arriving at night or at weekends are the most defensible extras.
- Using revenue instead of margin. A $1,200 job is not $1,200 of gain if $700 of it is materials and labour.
- Assuming follow-up happens. Captured enquiries that nobody rings back convert at zero. If your follow-up is slow, fix that first; why instant follow-up wins covers why it matters so much.
- Ignoring review time. On a $20 plan, 25 minutes a week of your time costs more than the software. Leaving it out makes every plan look like it pays back in days.
- Counting time savings you never reuse. Fewer repetitive emails is pleasant. It is only money if those hours go into quoting, fitting or selling.
Measuring the real figure after launch
Estimates get you to a decision. After launch, replace them with data. The analytics in SpideyChat show conversations, leads, visitors and unanswered questions, which covers steps 4 and part of 5. Your job book covers the rest.
Track two numbers each month: leads captured outside working hours, and leads from visitors who never touched the contact form. Tag which of those became paid work. After three months you have a payback period built on your own history instead of hopeful guesses. The metrics worth tracking are laid out in the metrics that prove your chatbot is generating leads.
Run your own numbers today
Copy the three-scenario table into a spreadsheet and fill the pessimistic column first, using your real hourly value, conversion rate and margin. If it pays back within the months you are prepared to commit, start on Free or Starter and track the two monthly numbers above. If it does not, the chatbot ROI calculator is a quick way to see which input would have to change, and whether that change is realistic for your business.