AI agent ROI calculator

The formula behind the calculator

The calculator on this page uses one core formula. It estimates how much new business you could recover by answering enquiries you currently miss, then subtracts what the AI agent costs. It deliberately leaves out harder-to-measure benefits, such as staff time saved or better customer experience, so the headline number stays grounded. You can think about those separately.

Each input is something you can estimate from your own records. Missed enquiries come from call logs, voicemails, unanswered messages and forms that waited too long. Conversion rate is the share of enquiries that normally become paying customers. Average value is what a typical new customer spends, either on the first job or over a year if you prefer. Cost is the monthly plan price plus any expected overage.

The calculator works on a monthly basis because that matches how most AI agent plans are billed. If you think in weeks, multiply weekly missed enquiries by about four and a third. If your business is seasonal, run the numbers for a quiet month and a busy month separately. A plan that pays for itself only in peak season may still be worth it, but you should know that going in.

• Recovered revenue = missed enquiries per month × conversion rate × average customer value
• Net monthly value = recovered revenue − monthly cost of the AI agent
• Return multiple = net monthly value ÷ monthly cost

A worked example (hypothetical)

The numbers below are invented for illustration, not benchmarks. Suppose a small plumbing business misses about five calls a week while the owner is on jobs, roughly 20 a month. Suppose one in four of those callers would have booked if someone answered, a conversion rate of 25 percent. That is five extra jobs a month. If an average job is worth $300, recovered revenue is 5 × $300 = $1,500 a month.

Now subtract the cost. On a plan at $99 a month with enough minutes included for this volume, the net monthly value is $1,500 − $99 = $1,401. Dividing by the cost gives a return multiple of about 14, meaning each dollar spent returns roughly fourteen in estimated new revenue. Revenue is not profit, so if you want a stricter view, use your margin per job instead of the job value.

Change one input and the picture changes. If only one in ten missed callers would have booked, recovered revenue drops to 2 jobs, or $600, and net value to $451 a month. If average job value is $1,200, as it might be for a roofing or HVAC repair, the same 25 percent conversion produces $6,000. That sensitivity is why it is worth spending a few minutes getting realistic inputs.

Getting realistic inputs

Missed enquiries are often underestimated. Count calls that rang out, went to voicemail, or were answered so late the caller had moved on; messages answered the next day; and website forms that sat over a weekend. If you do not have logs, track two typical weeks. Include after-hours contacts, since those are the ones an AI agent most obviously catches.

For conversion rate, start with your normal rate for enquiries you do answer, then be more conservative for missed ones, since some missed callers will try again or were never serious. If you are unsure, test a low, medium and high value in the calculator. For average value, decide whether to use first-job value or a year of repeat business, and be consistent. Using first-job value gives a cautious estimate.

Cost should reflect your realistic volume. Clientager plans start from $39 a month with 200 voice minutes, with larger plans for higher volumes and a per-minute rate above the allowance. Setup is included. See pricing for current plans, as prices may vary by region. If your volume is close to a plan's limit, run the calculator with the next plan up as well, so a busy month does not catch you out.

Benefits the formula leaves out

The core formula only counts new customers from missed enquiries. An AI agent also handles enquiries you would have answered anyway, which saves time. If your team spends hours each week on calls about hours, prices, rescheduling and order status, estimate those hours and multiply by an hourly cost. That figure is a separate, internal benefit that can be as meaningful as new revenue.

Other benefits are real but harder to price: fewer no-shows thanks to reminders, more repeat business from follow-ups, quotes that close because someone checked in, and an owner who can finish a job without picking up the phone. Treat these as upside on top of a conservative core estimate rather than adding them all to the headline number.

There are also costs beyond the plan. Allow for an hour or two a month reviewing conversations and requesting updates, and for any changes needed to your phone setup. Include reminders and follow-ups only to customers who have agreed to receive them, since outbound messages are subject to consent rules, and make sure the agent discloses that it is AI, as the EU AI Act requires from August 2026.

Using the result to make a decision

If the estimate shows a clear positive return even with cautious inputs, the decision is usually straightforward. If it only works with optimistic inputs, start smaller: one channel, the lowest plan, and a short review after the first month. Compare the agent's actual logged outcomes, such as bookings made and leads captured, with your estimate, then adjust.

Remember that the calculator gives an estimate, not a promise. Real results depend on your market, your pricing and how well you follow up the leads the agent captures. The useful thing about an AI agent is that every conversation is recorded, so after a month you can replace assumptions with your own data and recalculate with confidence.

It is also worth sharing the result with anyone else involved in the decision, such as a business partner or office manager. Walk them through the inputs rather than just the final number. Agreeing on realistic assumptions upfront makes it much easier to judge the outcome later, and avoids a debate about whether the agent worked when the real question was whether the estimate was sensible.

Common mistakes when estimating ROI

The most common mistake is assuming every missed enquiry would have become a customer. Some callers were price shopping, some needed a service you do not offer, and some would have called back anyway. Applying a realistic conversion rate, lower than for enquiries you answer promptly, keeps the estimate honest. The second mistake is using an unusually large job as the average value, which inflates every result.

A third mistake is ignoring costs beyond the plan price. Overage in busy months, your time reviewing conversations, and any changes to your phone setup all count. A fourth is counting the same benefit twice, for example adding both the revenue from recovered leads and the time saved answering those same calls. Keep revenue from missed enquiries and time saved on routine work as separate figures.

The final mistake is never checking the estimate against reality. Once the agent is live, compare its logged results with your assumptions after a month. If bookings from missed calls are higher than you expected, you have a stronger case to expand to another channel. If they are lower, look at why: the wrong channel, the wrong hours, or follow-up that is not happening on your side.