Calculating the ROI of an AI chatbot before commissioning a build requires three numbers: what the process currently costs your business, what percentage of that work the chatbot will realistically handle, and what the chatbot will cost to build and run. This guide gives you the formula, the realistic benchmarks for each variable, and worked examples so you can build a credible business case before spending anything.
Annual ROI = (Annual Cost Saved) minus (Annual Total Cost of Chatbot) divided by (Annual Total Cost of Chatbot) multiplied by 100.
Payback Period (months) = Total Build Cost divided by Monthly Cost Saved.
Both calculations are straightforward once you have the right inputs. The difficulty is calculating the inputs honestly rather than optimistically.
Calculate what the chatbot's target process currently costs your business per year. For a customer support chatbot, this means the total cost of handling customer queries through your current channel (email, phone, live chat, or a combination).
Start with the number of queries handled per month. Multiply by the average time per query in minutes. Divide by 60 to get hours. Multiply by the hourly cost of the agent handling queries (annual salary divided by 1,800 working hours, plus employer NI at 13.8% on earnings above the secondary threshold, plus a 20% to 25% overhead factor for workspace, equipment, and management). Multiply by 12 for the annual total.
Example: A London e-commerce business handles 800 support queries per month. Average handling time is eight minutes per query. Agents earn Β£28,000 per year (Β£15.56 per hour at 1,800 hours). With employer NI and overhead, the true cost per hour is approximately Β£20.00. Monthly query handling cost: (800 multiplied by 8 divided by 60) multiplied by Β£20.00 = Β£2,133. Annual cost: Β£25,600.
The automation rate is the percentage of queries the chatbot will resolve without human intervention. This varies significantly by query type and by how well the chatbot is trained and maintained. Use these benchmarks as your starting point.
Always use the conservative end of the range in your business case. It is better to exceed a conservative projection than to miss an optimistic one.
Chatbot costs break into three categories: build cost, ongoing running cost, and maintenance cost.
Using the e-commerce business from the example above:
Current annual cost: Β£25,600 (800 queries per month at Β£20 per hour).
Automation rate applied: 65% (conservative estimate for a mixed-query retail support chatbot).
Annual cost saved: Β£25,600 multiplied by 65% = Β£16,640.
Build cost: Β£25,000 (custom chatbot with order management integration).
Annual running cost: Β£7,200 (Β£400 platform and API, Β£200 hosting, per month multiplied by 12).
Annual maintenance: Β£4,000 (16% of build cost).
Total first-year cost: Β£25,000 build plus Β£7,200 running plus Β£4,000 maintenance = Β£36,200.
First-year ROI: (Β£16,640 minus Β£36,200) divided by Β£36,200 = minus 54%. Negative in year one due to the build cost.
Annual cost from year two: Β£7,200 running plus Β£4,000 maintenance = Β£11,200.
Year two ROI: (Β£16,640 minus Β£11,200) divided by Β£11,200 = 48.6%.
Payback period: Β£25,000 build cost divided by (Β£16,640 minus Β£11,200 annual saving net of running costs) = 4.6 months of net saving to recover the build cost. Total payback from launch: approximately 22 months.
This example shows a solid long-term ROI with a 22-month payback period. Whether that meets your hurdle rate depends on your business's capital allocation policy.
Query volume: Higher volume makes the ROI calculation more favourable because the running cost is largely fixed while savings scale with volume. A business handling 2,000 queries per month sees the same payback period improve to under 12 months with the same build cost.
Agent cost: Businesses with higher-cost agents (London salaries, specialist knowledge requirements) see faster payback than businesses with lower-cost agents.
Automation rate: A 10% increase in automation rate from 65% to 75% on 800 queries per month adds approximately Β£2,560 per year in additional savings. Investing in better training data and knowledge base quality to improve automation rate is often the highest-ROI activity after initial deployment.
For businesses handling more than 500 support queries per month with an average query cost above Β£8, most AI chatbot deployments pay back the build cost within 12 to 24 months. High-volume businesses (2,000+ queries per month) with higher agent costs often see payback within six to twelve months. Low-volume businesses (under 200 queries per month) rarely achieve payback within two years and should evaluate whether an AI chatbot is the right investment.
Three costs are consistently underestimated: the time required to build and maintain the knowledge base that powers the chatbot, the cost of handling escalated queries from the chatbot (human agents handling escalations need context and the chatbot needs to transfer it effectively), and ongoing maintenance costs as your products, policies, and integrated systems change over time. Build all three into your calculation from the start.
To get a specific ROI estimate for your business based on your actual query volume and costs, see our AI Chatbot Development service for London businesses.
More from the blog
Let us help
Talk to our London-based team about how we can build the AI software, automation, or bespoke development tailored to your needs.
Softomate Solutions Limited
Deen Dayal Yadav
Online
