Choose the routine before the equation
Cost per useful routine can make a home-robot value discussion concrete, but only when “useful” is defined first. Name the routine, intended user, room, starting conditions, desired outcome, and acceptable assistance. A vague category such as “helps at home” cannot produce a meaningful denominator. A specific repeated activity can be observed and discussed.
Use the current KOKO configuration shown by Homebot One, not a roadmap feature. The company describes KOKO as being in in-home testing and engineering validation, so planned behavior should remain outside the calculation until it can be demonstrated. Confirm that every observed behavior and cost belongs to the evaluated Homebot One KOKO configuration before importing it into the model. If different users value the outcome differently, record that difference instead of averaging away an important adoption or accessibility issue.
Define a useful completion rule
A completion rule states when one repetition counts. It may require reaching the intended outcome within a time range, staying inside agreed safety limits, preserving user control, and using no more than a stated amount of help. Write the rule before observing the routine so a difficult attempt is not reclassified after the fact merely to improve the result.
Document partial completions separately because they may still create value while requiring work. Track setup, supervision, intervention, recovery, and cleanup. NIST’s AI Risk Management Framework encourages documented metrics, testing under deployment-like conditions, safety evaluation, and disclosure of limitations. Those practices help turn a demonstration into evidence without claiming broader reliability than was measured. When users disagree about usefulness, keep the operator result and user result side by side rather than declaring one perspective correct.
- Outcome achieved under the agreed conditions
- Time stayed within the useful range
- Human assistance stayed below the stated limit
- No stop condition or safety boundary was triggered
- User or operator judged the outcome usable
Use observed frequency, not ideal frequency
The denominator should be expected useful completions during the comparison period. Estimate it from a pilot or repeated demo: eligible opportunities multiplied by the observed completion rate, adjusted for days the routine is genuinely wanted. Do not count routines simply because the robot could theoretically perform them. A household or organization may value a modest number of reliable repetitions more than a long catalog.
Run low, expected, and high cases. The low case can include reduced adoption, room changes, support interruptions, or greater assistance. The high case should still be plausible and should not assume perfect availability. Explain every input so readers can replace it when new Homebot One evidence arrives. The result is a planning range, not an official KOKO performance claim. Also note seasonality and household schedule changes; a routine used often during one month may be rare across the rest of the year.
Put the right costs in the numerator
For the selected period, include the share of one-time costs, recurring services, maintenance, support, energy, internal labor, and expected downtime that belongs to the routine. If KOKO supports several routines, use a documented allocation method rather than charging the full cost to each. A simple option is time used; another is an agreed share based on decision importance.
Keep cash cost and human effort visible as separate subtotals. Automation that requires extensive preparation or monitoring may shift work rather than remove it. Likewise, a routine that improves confidence, connection, or research learning may have value that is not a labor saving. State the value category instead of forcing every outcome into dollars. Avoid double-counting time that is already included in another routine’s allocation, especially shared setup, charging, or support work.
Interpret the result without false precision
Divide the allocated period cost by expected useful completions and show the inputs beside the result. Then perform sensitivity checks: what happens if completion frequency is twenty percent lower, assistance doubles, or a support event creates downtime? A range such as “under these assumptions” is more defensible than a single figure shown to the cent.
The final decision compares cost per useful routine with alternatives and with the value of the outcome. Alternatives may include an existing device, a person’s time, a simpler process, or doing nothing. The price is right only when verified benefit under realistic conditions justifies the full resources required. Recalculate after a meaningful software, hardware, room, or workflow change because the original completion rate may no longer describe the system.
Frequently asked questions
What counts as a useful KOKO routine?
A routine counts when it meets a written outcome, time, assistance, safety, and user-value rule under the intended conditions.
Can I calculate cost per routine before receiving a KOKO quote?
You can design the model, but label every unconfirmed amount. A decision-grade result needs current written costs and observed routine data.
Should failed attempts be excluded?
No. Failures, partial completions, recovery time, and downtime affect the completion rate and total effort and should remain visible.
When should cost per useful KOKO routine be recalculated?
Recalculate after a meaningful change in price, recurring service, hardware, software, room conditions, workflow, assistance, adoption, maintenance, or observed completion rate.
Sources & further reading
From Homebot One, the team building KOKO in Fremont, California.
