cloud phone TCO worksheet: real numbers for 2026
cloud phone TCO worksheet: real numbers for 2026
a cloud phone TCO worksheet in 2026 should answer one question: what does this actually cost over three years, with all the line items the sales deck did not mention. most TCO calculations stop at “device price plus subscription”. real TCO includes engineer time, network egress, screenshot storage, replacement cycles, off-boarding, and the hidden tax of whatever the vendor invoices for in year two that they did not in year one. this worksheet covers all of it.
if you have read the cloud phone vs physical lab TCO breakdown, this is the spreadsheet you build after. for the broader buying flow, the vendor checklist and RFP template come earlier.
the worksheet structure
four blocks: setup, recurring, hidden, exit. each block has line items you fill from your vendor’s quote and your own internal numbers.
block 1 = year zero (one-time) blocks 2-3 = annual operating block 4 = end-of-contract or migration
sum across three years to get TCO. compare across vendors at three scale points (10, 50, 100, 500 phones).
block 1: setup costs (year zero)
| line item | typical range USD | source |
|---|---|---|
| onboarding fee | 0 to 5,000 | vendor quote |
| integration engineering time | 40-160 hours | internal estimate |
| security review time | 8-24 hours | internal estimate |
| legal review time | 4-16 hours | internal estimate |
| training time (per seat) | 2-4 hours | internal estimate |
| internal documentation | 8-16 hours | internal estimate |
| pilot APK signing infra | 500-2,000 | one-time |
multiply engineer hours by your loaded cost (typically $100-200/hour US, $50-100/hour SG/EU). a fully loaded onboarding for a 50-phone deployment is usually $15-40k of internal time on top of any vendor onboarding fee.
block 2: recurring costs (per year)
| line item | typical range | how to compute |
|---|---|---|
| device subscription | $30-120/device/month | vendor pricing tier |
| seat licenses | $0-50/seat/month | vendor pricing |
| compute / minutes overage | varies | track actual minutes from POC |
| screenshot/recording storage | $5-30/100GB/month | vendor pricing |
| network egress | $0.05-0.15/GB | vendor pricing or cloud bill |
| dedicated IP / proxy add-on | $5-20/IP/month | vendor pricing |
| support tier upgrade | 5-15% of subscription | vendor quote |
| optional residential proxy | $5-30/GB | external proxy provider |
annual subscription for 50 phones at $60/device/month is $36k. typical realistic all-in number once you add storage, egress, and support tier is around $42-48k.
block 3: hidden / variable costs
these are the line items vendors do not put in the sales deck.
| line item | likely cost | note |
|---|---|---|
| ops engineer time | 5-15% FTE per 50 phones | provisioning, debugging, vendor liaison |
| in-house dashboard or wrapper | 60-200 hours | building org-internal tooling on top of vendor API |
| audit-log shipping / SIEM | $1-3k/year | sending logs to Splunk/Datadog |
| compliance reporting | 8-40 hours/year | per audit cycle |
| developer wait time on flaky tests | varies | track via CI metrics |
| price increase on renewal | 5-15% YoY | per MSA cap |
| device drift cost | 2-8% test re-baseline/year | when vendor refreshes models |
| integration drift cost | varies | when vendor breaks API |
ops engineer time is the biggest one. teams systematically underestimate it. budget at least 10% of one engineer’s time per 50 phones in production for the first year. it drops to 5% by year two if you build automation.
block 4: exit / migration costs
every contract ends. plan for the cost.
| line item | typical | source |
|---|---|---|
| data export fee | 0-5,000 | MSA |
| migration engineering time | 80-240 hours | internal estimate |
| parallel-run period | 1-3 months of dual-vendor cost | internal estimate |
| audit log archival | $500-2,000 | retention requirement |
| training for new vendor | 16-40 hours per seat | internal estimate |
at 50 phones, a migration to a new vendor typically costs $30-80k all in, even if the new vendor is cheaper.
three-year totals at four scale points
assumes mid-range vendor pricing, US engineer loaded cost.
| scale | year 1 | year 2 | year 3 | 3-yr TCO |
|---|---|---|---|---|
| 10 phones | $22k | $14k | $14k | $50k |
| 50 phones | $90k | $62k | $66k | $218k |
| 100 phones | $165k | $115k | $122k | $402k |
| 500 phones | $720k | $520k | $555k | $1.79M |
these include block 3 hidden costs at realistic levels. real numbers will swing $20-30% depending on vendor, region, and team maturity.
comparing two vendors
run the same worksheet for vendor A and vendor B. then compute three numbers per scale point.
- 3-year TCO (vendor A vs B)
- year 1 TCO (cash-flow sensitive teams care about this)
- TCO sensitivity to scale (does one vendor become much cheaper at 100+?)
a useful tiebreaker question: at what scale does vendor A become cheaper than vendor B, and how confident are we that we will hit that scale? if the answer is “we save money once we cross 200 phones and we are at 30 today,” that is not real savings.
sensitivity analysis
four variables that swing TCO by more than 20%.
- price increase cap on renewal. 5% vs 15% per year compounds fast.
- engineer hourly rate. $100/hr vs $200/hr changes hidden costs.
- utilization rate. 60% vs 90% device utilization changes effective $/device-hour by 50%.
- screenshot/recording retention. keeping forever vs 30 days changes storage 10x.
build a one-page sensitivity table that flips each variable up and down. the lowest-TCO vendor at the median assumption may not be the lowest at your actual numbers.
what the worksheet usually reveals
three patterns repeat.
- vendor advertised pricing is 60-70% of real first-year cost. the rest is integration, ops, security review, hidden line items.
- vendor B always wins on advertised pricing. vendor A often wins on three-year TCO once support tier, hidden costs, and renewal cap factor in.
- the cheapest provider is rarely the cheapest provider. the lowest sticker price often correlates with weak support and high engineer-hour overhead.
run the worksheet anyway. the wrong answer is “no vendor is cheap” with no model behind it.
frequently asked questions
how do I model engineer hours when I do not yet know the integration depth?
start with three buckets: integration (40h baseline + 20h per major CI integration), ops (5% of one FTE per 50 phones), incident response (2h per P1, expect 4-12 P1s per year per 100 phones).
what about cloud egress costs?
if your CI runners are in AWS/GCP/Azure and the cloud phones are in another region, egress is $0.05-0.15 per GB. for typical mobile testing (logs + screenshots), this is $1-5 per device per month. small but not zero.
should I include opportunity cost of capex tied up in physical phones?
if you are comparing cloud phone vs physical lab, yes. capex tied up in $80k of devices that will be obsolete in 30 months is a real cost. cloud phone subscription is opex with no residual value but no obsolescence risk either.
how do I handle multi-region pricing?
most vendors price per region or charge a regional premium of 10-30%. add it as a separate line item per region you actually use.
what is a reasonable contingency for hidden costs in year 1?
15-20%. teams that budget 10% almost always overrun. teams that budget 25% leave money on the table.
ready to fill the worksheet against a real vendor? start a cloudf.one trial and use the actual usage data from days 1-14 as your input numbers. far more accurate than any vendor quote.