AI deflection ROI calculator
Vendor calculators multiply contacts by a deflection rate and call the result savings. Two things are missing: deflected contacts that were not resolved come back, and agents do not scale in a straight line with volume. This calculator separates deflection from containment, returns the failures to the queue, and runs the peak hour through Erlang C, then prices the difference against what the bot costs.
After the bot's own cost.
- FTE saved
- 8.1
- Agents saved at the peak hour
- 5
- Contacts removed per day
- 495
- Contacts that come back
- 405
- Vendor-style estimate
- £537,815
Erlang C, 80/20.
Deflected and contained.
Deflected, not contained.
Deflection × handle time, no returns, no bot cost.
Show the working
- Deflected: 3000 × 30% = 900 contacts a day. Contained: × 55% = 495 removed; 405 come back to an agent at 300 s × 1.15.
- Handling hours a day fall from 250.0 to 213.8. At 85.0% occupancy and 30% shrinkage an FTE delivers 1160 handling hours a year, so FTE falls from 56.0 to 47.9: 8.1 saved.
- Peak hour: 375 contacts need 37 agents for 80/20; after deflection 313 contacts at 307 s need 32.
- Gross saving 8.1 × £32000 = £259496 a year; less the bot at £60000 = £199496 net. A vendor-style linear estimate (deflection alone, no returns) would claim £537815.
Doing this for every interval of the week? Pebble WFM computes the requirement from your forecast and builds the roster. Free month, no card needed.
How the deflection saving is calculated
Deflection is the share of contacts the bot takes first. Containment is the share of those it actually resolves, measured as no follow-up on any channel. The difference comes back to an agent, usually with a longer handle time because the customer has already tried once. Only the contained contacts are removed from the workload.
The reduced workload is converted to FTE through occupancy and shrinkage, which is how the roster actually shrinks. The peak hour is also run through Erlang C before and after, because the agents needed at the peak do not fall in proportion to volume: a smaller queue needs proportionally more slack.
The net figure subtracts the bot's annual cost. The vendor-style line beside it is what a linear calculator would claim from the same inputs, and the gap between the two is the number to take into the procurement conversation.
Frequently asked questions
- What is the difference between deflection and containment?
- Deflection counts sessions the bot handled. Containment counts customers who did not come back. A bot can deflect 40 per cent and contain 20; the other 20 phoned. Only containment saves agent time, and it is the number vendors rarely quote.
- Why does the peak-hour saving differ from the FTE saving?
- FTE is a throughput number over the year. The peak hour is a queueing number: fewer contacts need fewer agents, but the slack for random arrivals grows as a share, so the saving at the peak is smaller than the volume drop suggests. Rosters are built to the peak.
- How do I measure containment properly?
- Take a sample of deflected sessions and check every channel for a contact from the same customer about the same issue within a day or two. Bot analytics cannot see the phone call that followed the chat.
Embed this calculator on your site
Paste this where you want the calculator. It stays free and ad-free, updates when we improve it, and the line under it credits the source.
Stop doing this one interval at a time
Pebble WFM forecasts your demand, computes the staffing requirement for every interval, builds the roster and publishes it, with self-service for agents and a copilot that can do what a planner can. Explore a sample organisation on day one.