Asato ROI Calculator
The Asato ROI Calculator estimates the potential yearly savings an organization could reach, grouped under the things Asato works on: AI spend, software, hardware, the application portfolio and contracts. It takes employees, AI spend, software spend and hardware spend, and works through a list of saving events, each one the value of one unit multiplied by an estimated whole number of units a year. The result is a range with an expected value. It is an estimate: not a quote, not a guarantee and not a customer result.
When to use
- A buyer wants a first, order-of-magnitude view of potential savings before talking to Asato.
- A buyer wants to see how such an estimate is built: which part of each figure is a published price and which part is an estimate.
- An agent needs to reproduce the estimate for a given organization. Everything needed is on this page.
Not a fit
- A business case that has to survive audit. The calculator has not seen the organization's estate. The Asset Clarity Report prices one area of the real estate in one week.
- AI billed by use, through a meter or an API. No neutral figure exists for how much of it could be recovered, so the calculator has no line for it. The AI lines cover assistant seats only.
- An organization's own counts. The page has no field for them: every count is the model's.
How it works
Inputs
The page arrives filled in for 5,000 employees. Size buttons (500, 5,000 and 25,000 employees) set the number of employees and return the spend fields to their defaults.
| Input | Default | What it drives |
|---|---|---|
| Employees | 5,000 | The lines that follow headcount, and the spend defaults |
| AI spend, a year | $108 per employee ($540,000 at 5,000 employees) | Assistant seats = AI spend / $360. The AI lines follow the seats |
| Software spend, a year | $1,416 per employee ($7,080,000 at 5,000 employees) | The software license lines. Below its default, it also scales down the leaver, portfolio and renewal lines that draw on it |
| Hardware spend, a year | $390 per employee ($1,950,000 at 5,000 employees) | The device lines. Below its default, it also scales down the leaver, server, maintenance and cloud lines that draw on it |
Each spend field is a default the visitor can overwrite. Once a field is overwritten, "Use the default" appears beside it and returns it to the default, and emptying the field does the same.
- What a field reads. A full figure with or without separators ("40,000,000"), a decimal point, and K, M or B for thousands, millions and billions ("40M", "7.5M", "1.2B"). The field then shows the full figure with separators. Amounts are whole dollars.
- What a field cannot read. Anything else leaves the last good value in place, and a line under the field says so while the field has focus.
- Limits. Employees takes 1 to 9,999,999. Each spend field takes up to $99,999,999,999.
- Calibration. The model is calibrated from 500 to 50,000 employees. Outside that range the page says so under the Employees field.
- AI default. An Asato assumption: 30% of employees hold one paid assistant seat at $360 a year (the list price of Microsoft 365 Copilot), which is $108 per employee.
- Hardware default. 260 business-class laptops bought a year per 1,000 employees (1.05 devices per employee on a four-year refresh) at $1,500 each: $390 per employee.
- Software default. The cost, at list price, of the seats the model assumes per 1,000 employees:
| Seats per 1,000 employees | Product | List price | A year |
|---|---|---|---|
| 380 | Microsoft 365 E5 | $60 a month | $273,600 |
| 570 | Microsoft 365 E3 | $39 a month | $266,760 |
| 36 | Salesforce Sales Cloud Advanced | $395 a month | $170,640 |
| 84 | Salesforce Sales Cloud Core | $195 a month | $196,560 |
| 30 | Adobe Creative Cloud Pro, all apps | $99.99 a month | $35,996 |
| 1,100 | Collaboration seats (Slack, Box, Tableau, Figma) | $32.50 a month, the average of four list prices | $429,000 |
| 80 | Docusign Business Pro | $45 a month | $43,200 |
| Total per 1,000 employees | $1,415,756.40, which is $1,416 per employee after rounding |
The calculation
For every line: potential savings a year = whole units a year x the value of one unit. Units a year follow one of these rules, given for each line in the tables below:
- AI spend. Units are a share of assistant seats, and seats = AI spend / $360.
- Software spend. Units = the line's rate per 1,000 employees x software spend / $1,416,000. With the default spend this equals the rate x employees / 1,000.
- Hardware spend. Units = the rate per 1,000 employees x hardware spend / $390,000.
- Leavers. Units = the rate per 1,000 employees x employees / 1,000. The model assumes 13% of staff leave in a year.
- A power of headcount. Units = units at 5,000 employees x (employees / 5,000) to a power below 1: 0.35 for portfolio and renewal lines, 0.7 for infrastructure lines.
- The spend a headcount line draws on. Each line under the last two rules draws on software spend or on hardware spend, named in its rule in the tables below. At the model's default spend for the headcount, or above it, the rule stands as written. Below it, the units are multiplied by that spend / its default, so the line shrinks as the spend does and counts nothing at no spend. With all three spend fields at zero, every line is zero and the estimate is $0.
Units are whole. The count a rule gives is rounded to the nearest whole unit, a half rounded up, and that whole number is what the page shows. So each line's value is its shown units times its shown price, each solution is the sum of its lines, and each tier and the total are sums of those. Every figure can be checked by hand.
When a spend field has been overwritten, the page states a line's count as a multiple of the model's default spend for the headcount, and never as the seats or devices of a number of employees. Example, for 12,000 employees with $40,000,000 of software spend:
35 a year per 1,000 employees is 420 at 12,000 employees. Your software spend is 2.35 times the model's $16,992,000 for that headcount, so the count scales to 989.
A visitor can switch any priced line, or a whole solution, off. The one line that is counted and not priced has no switch, because it adds nothing to the estimate.
The range
The price basis grades the value of one unit:
- Published price. The value of one unit is a vendor's own list price.
- Part published. Part of the value is published and part is an Asato estimate.
- Asato estimate. Built from structure, with no published figure behind it.
- Counted, not priced. A count with no dollar value.
The estimate is a range built from those grades:
- Low = the sum of the lines with a published price.
- Expected = low + the lines that are part published.
- High = expected + the lines priced by Asato estimate.
How often each event happens (units a year) is an Asato estimate on every line, so even the low end is a published price multiplied by an estimated frequency and is not a verified figure.
What the page shows
- The result. Under the label "Estimated potential savings a year, for N employees", the expected value as one large figure, then the three tiers, then the expected value per employee. Summary figures are shown to three significant figures (for example $891K and $1.30M). Line values and the sum under each solution are shown to the dollar.
- The result in view. Once the large figure scrolls up to the top of the window, a thin strip there carries the expected value and its range for as long as the calculator is on screen. The strip sits under the navigation bar whenever the bar is showing.
- The offer, beside the result. "Your own estate, priced in one week. If it finds no savings, you pay nothing." and the button "Get your Asset Clarity Report".
- Five solutions. Each opens to its lines. A line shows its whole units, the value of one unit and their product, and opens to its unit, its count and the basis of that count, its price, its trigger and its linked sources.
- Motion. Figures ease to a new value and the bars re-scale when an input or a switch changes. With reduced motion set, values simply change.
A link to an estimate
"Copy a link to this estimate" writes the page's address, with the present values after the hash, to the visitor's clipboard. If the browser refuses, the page shows the address as selected text to copy by hand. The address is built only when the visitor asks. The page reads the values when it opens and then takes them out of the address bar.
Form: #n-EMPLOYEES-SOFTWARE-HARDWARE-AI, then optionally -off. and the ids of the lines and solutions that are switched off, joined by dots. Each spend is in millions of dollars with up to six decimals, or d for the default. Only letters, digits, dots and hyphens are used.
#n-12000-40-15: employees, software spend and hardware spend written into the address, the spends in millions. 12,000 employees, $40,000,000 of software spend, $15,000,000 of hardware spend, AI spend at its default.#n-8000-25-9-d: 8,000 employees, $25,000,000 of software spend, $9,000,000 of hardware spend, AI spend at its default.#n-5000-d-d-1.25-off.e12.pf: 5,000 employees, software and hardware spend at their defaults, $1,250,000 of AI spend, with the line e12 and the solution pf switched off.
Solution ids: ai Control AI spend, sw Right-size software, hw Stop paying for idle hardware, pf Rationalize your portfolio, ct Stay ahead of contracts. Line ids are in the table of every line below.
The saving lines, by solution
Control AI spend
Solution page: https://www.asato.ai/control-ai-spend
| Line | Value of one unit | Price basis | Units a year, before rounding to a whole unit | Estate it draws from | Share of estate |
|---|---|---|---|---|---|
| Idle AI assistant seat reclaimed | $360 | Published price | 10% of assistant seats, where seats = AI spend / $360 | assistant seats | 10.0% |
| Overlapping AI assistant removed | $228 | Part published | 5% of assistant seats, where seats = AI spend / $360 | assistant seats | 5.0% |
| AI seat released on the leaver date | $30 | Part published | 13% of assistant seats, where seats = AI spend / $360 | assistant seats | 13.0% |
Idle AI assistant seat reclaimed. One paid assistant seat taken back from a person who does not use it.
- Value of one unit (published price): Microsoft 365 Copilot lists at $30 per user a month, paid yearly: $360 a year.
- How many (Asato estimate): 10% of seats: an Asato estimate, set at half of what a randomized trial measured. Across 66 firms and 7,137 workers in 2023 and 2024, 20% of workers given an assistant never used it, and weekly use settled just under 40%.
- What triggers it: A paid assistant seat with no recorded use.
- Sources: Microsoft 365 Copilot pricing; Dillon and co-authors, NBER working paper 33795: a randomized trial of an AI assistant
Overlapping AI assistant removed. One second assistant seat removed where the person already holds another.
- Value of one unit (part published): GitHub Copilot Business lists at $19 a seat a month: $228 a year. The price is published. Which assistant is the second one is an Asato estimate, so the lower of the two list prices used here stands in for it.
- How many (Asato estimate): 5% of seat holders hold a second assistant: an Asato estimate. For context, the Ramp AI Index reports that in April 2026, among businesses paying by card or invoice through Ramp, 50.6% paid for AI, 34.4% paid Anthropic and 32.3% paid OpenAI. The two vendor shares add up to 66.7%, more than the total, so by arithmetic at least 16% of those businesses paid both. That counts businesses with two vendors, not people with two seats. In a June 2025 a16z survey of 100 CIOs, 37% said they use five or more models.
- What triggers it: One person holds paid seats on two assistants.
- Sources: GitHub Copilot plans; Ramp Economics Lab, Ramp AI Index, May 13, 2026; a16z, AI in the enterprise, June 2025
AI seat released on the leaver date. One leaver's assistant seat released a month earlier than it otherwise would be.
- Value of one unit (part published): One month of a $30 seat. The price is published. That the seat would otherwise stay assigned for a month is an Asato estimate.
- How many (Asato estimate): 13% of seats, on the model's assumption that 13% of staff leave in a year. That is conservative: US Bureau of Labor Statistics data for 2025 put total separations at 3.3% a month and quits at 2.0% a month, about 40% and 24% a year.
- What triggers it: A leaver event in the HR system while an assistant seat is still assigned.
- Sources: Microsoft 365 Copilot pricing; US Bureau of Labor Statistics, Job Openings and Labor Turnover, released March 13, 2026
Right-size software
Solution page: https://www.asato.ai/right-size-software
| Line | Value of one unit | Price basis | Units a year, before rounding to a whole unit | Estate it draws from | Share of estate |
|---|---|---|---|---|---|
| Microsoft 365 E5 moved down to E3 | $252 | Published price | 35 x software spend / $1,416,000 | 380 Microsoft 365 E5 seats per 1,000 employees | 9.2% |
| Microsoft 365 seat reclaimed | $468 | Published price | 40 x software spend / $1,416,000 | 950 Microsoft 365 seats per 1,000 employees | 4.2% |
| Adobe Creative Cloud reduced to a single app | $744 | Published price | 6 x software spend / $1,416,000 | 30 Adobe All Apps licenses per 1,000 employees | 20.0% |
| Salesforce seat reclaimed | $2,340 | Published price | 8 x software spend / $1,416,000 | 120 Salesforce seats per 1,000 employees | 6.7% |
| Salesforce edition moved down | $2,400 | Published price | 4 x software spend / $1,416,000 | 36 Salesforce Advanced seats per 1,000 employees | 11.1% |
| Duplicate Power BI Pro license removed | $168 | Published price | 12 x software spend / $1,416,000 | 380 Microsoft 365 E5 seats per 1,000 employees | 3.2% |
| Collaboration seat reclaimed | $390 | Published price | 55 x software spend / $1,416,000 | 1,100 collaboration seats per 1,000 employees | 5.0% |
| Docusign plan right-sized | $180 | Part published | 8 x software spend / $1,416,000 | 80 Docusign seats per 1,000 employees | 10.0% |
| Licenses released on the leaver date | $96 | Asato estimate | 130 x employees / 1,000, x (software spend / its default) when that spend is below its default | 130 leavers per 1,000 employees | 100.0% |
| Orphaned account or OAuth grant revoked | Not priced | Counted, not priced | 45 x employees / 1,000, x (software spend / its default) when that spend is below its default | 130 leavers per 1,000 employees | 34.6% |
Microsoft 365 E5 moved down to E3. One license moved down a tier, for a user who never touches the E5-only workloads.
- Value of one unit (published price): Microsoft list price from July 1, 2026: Microsoft 365 E5 is $60 per user a month and E3 is $39. The difference is $21 a month, $252 a year.
- How many (Asato estimate): 35 a year per 1,000 employees. No credible published rate of E5 over-provisioning exists, so this is a deliberately conservative Asato estimate. Estate: 95% of staff hold a Microsoft 365 license, and 40% of those are on E5.
- What triggers it: A user on E5 with no Power BI, Defender or advanced compliance activity in 90 days.
- Sources: Microsoft 365 pricing and packaging update, effective July 1, 2026
Microsoft 365 seat reclaimed. One E3 license unassigned and returned to the pool.
- Value of one unit (published price): Microsoft 365 E3 is $39 per user a month from July 1, 2026: $468 a year.
- How many (Asato estimate): 40 a year per 1,000 employees: an Asato estimate, since no neutral rate of unused seats exists. For context, the US Government Accountability Office reviewed nine federal agencies in 2024, and none had fully determined whether their five most widely used software licenses were over- or under-purchased. Estate: 95% of staff hold a Microsoft 365 license.
- What triggers it: No sign-in for 60 days.
- Sources: Microsoft 365 pricing and packaging update, effective July 1, 2026; US Government Accountability Office, GAO-24-105717, January 2024
Adobe Creative Cloud reduced to a single app. One All Apps license swapped for the one application the person opens.
- Value of one unit (published price): Adobe teams pricing: Creative Cloud Pro is $99.99 per license a month and a single app is $37.99. The difference is $62 a month, $744 a year.
- How many (Asato estimate): 6 a year per 1,000 employees: an Asato estimate. No independent study of Adobe utilization exists. The share looks high because the model assumes a small Adobe footprint. Estate: 3% of staff hold Creative Cloud All Apps.
- What triggers it: Only one Creative Cloud application launched in 90 days.
- Sources: Adobe Creative Cloud for teams pricing
Salesforce seat reclaimed. One Sales Cloud Core license unassigned.
- Value of one unit (published price): Salesforce Sales Cloud Core lists at $195 per user a month: $2,340 a year. Salesforce repackaged its editions in 2026 as Core, Advanced and Max.
- How many (Asato estimate): 8 a year per 1,000 employees: an Asato estimate. No independent study of Salesforce license waste exists. Estate: 12% of staff hold a Sales Cloud license.
- What triggers it: No login for 60 days, or logins with no record activity.
- Sources: Salesforce Sales Cloud pricing
Salesforce edition moved down. One license moved from Advanced to Core.
- Value of one unit (published price): Sales Cloud Advanced lists at $395 per user a month and Core at $195. The difference is $200 a month, $2,400 a year.
- How many (Asato estimate): 4 a year per 1,000 employees: an Asato estimate. Estate: 30% of Salesforce seats are on Advanced.
- What triggers it: A user on Advanced with no use of Advanced-only features in 90 days.
- Sources: Salesforce Sales Cloud pricing
Duplicate Power BI Pro license removed. One standalone Power BI Pro license removed where the user already has it through E5.
- Value of one unit (published price): Power BI Pro is $14 per user a month: $168 a year. It is included in Microsoft 365 E5, so a standalone license on an E5 user is duplication.
- How many (Asato estimate): 12 a year per 1,000 employees: an Asato estimate. The overlap is mechanical to detect once both records are held, and needs no usage data. Estate: the overlap can only exist on an E5 holder.
- What triggers it: A user holds both an E5 license and a standalone Power BI Pro license.
- Sources: Microsoft Power BI pricing; Microsoft 365 pricing and packaging update, effective July 1, 2026
Collaboration seat reclaimed. One Slack Business+, Box Enterprise, Tableau Enterprise or Figma Organization seat.
- Value of one unit (published price): The average of four 2026 list prices: Slack Business+ at $15 a month, Box Enterprise at $35, Tableau Enterprise at $35 and Figma Organization at $45, billed annually. That is $32.50 a month, $390 a year.
- How many (Asato estimate): 55 a year per 1,000 employees: an Asato estimate. For context, Gartner is reported to have said in 2016 that organizations can cut software spending by as much as 30%. That is a savings potential, not measured waste. Estate: about 1.1 seats per employee across Slack, Box, Tableau and Figma.
- What triggers it: No activity in 60 days.
- Sources: Slack pricing; Box pricing; Tableau pricing; Figma pricing; Gartner press release, July 2016, on cutting software costs (second-hand)
Docusign plan right-sized. One seat moved down a plan where the yearly envelope allowance goes unused.
- Value of one unit (part published): Docusign Business Pro is $45 per user a month and Standard is $30, a difference of $180 a year. Standard includes 100 envelopes per user a year. Both prices are published. That a low-volume seat can move to Standard is an Asato estimate.
- How many (Asato estimate): 8 a year per 1,000 employees: an Asato estimate. Docusign is priced on envelopes per user, so a count of active seats understates the waste. Estate: 8% of staff hold a Docusign seat.
- What triggers it: Envelope use far below the included allowance.
- Sources: Docusign eSignature pricing
Licenses released on the leaver date. One leaver's licenses released weeks earlier than they otherwise would be.
- Value of one unit (Asato estimate): An Asato estimate: three apps at an average of $400 a year each is $1,200. Releasing them 30 days earlier, about 8% of a year, is worth about $96. No published benchmark exists for the days between a leaver date and license release.
- How many (Asato estimate): One event per leaver, so this is every leaver by construction and not a reclaim rate. For context, in Capterra's 2023 Insider Threats Survey of 400 respondents, 36% of companies that reported insider attacks said a former employee with active credentials was involved. Estate: 13% of staff leave in a year.
- What triggers it: A leaver event in the HR system.
- Sources: Capterra, 2023 Insider Threats Survey; US Bureau of Labor Statistics, Job Openings and Labor Turnover, released March 13, 2026
Orphaned account or OAuth grant revoked. One access path closed that outlived the leaver process.
- Value of one unit (counted, not priced): Counted, not priced. IBM and Ponemon put the average breach at $4.99M globally and $11.5M in the United States, and report that 92% of organizations with an AI-related breach lacked AI access controls. Multiplying an average breach cost by a risk percentage gives a figure no CFO accepts, so this line stays a count.
- How many (Asato estimate): 45 a year per 1,000 employees: an Asato estimate. Estate: 13% of staff leave in a year.
- What triggers it: An active credential or OAuth grant that belongs to a terminated identity.
- Sources: IBM and Ponemon, Cost of a Data Breach 2026; US Bureau of Labor Statistics, Job Openings and Labor Turnover, released March 13, 2026
Stop paying for idle hardware
Solution page: https://www.asato.ai/idle-hardware
| Line | Value of one unit | Price basis | Units a year, before rounding to a whole unit | Estate it draws from | Share of estate |
|---|---|---|---|---|---|
| Idle device redeployed, not bought | $1,500 | Part published | 10 x hardware spend / $390,000 | 260 devices bought a year per 1,000 employees | 3.8% |
| Comatose server or VM decommissioned | $1,800 | Part published | 30 x (employees / 5,000) to the power 0.7, x (hardware spend / its default) when that spend is below its default | 600 server instances at 5,000 employees | 5.0% |
| Device recovered from a leaver | $1,500 | Part published | 18 x employees / 1,000, x (hardware spend / its default) when that spend is below its default | 130 leavers per 1,000 employees | 13.8% |
| Retired device sold for its residual value | $150 | Part published | 60 x hardware spend / $390,000 | 260 devices retired a year per 1,000 employees | 23.1% |
Idle device redeployed, not bought. One machine pulled from stock to fill a request.
- Value of one unit (part published): The avoided purchase of a business-class laptop: $1,200 to $1,900 configured, from retail listings. The model uses $1,500.
- How many (Asato estimate): 10 a year per 1,000 employees: an Asato estimate. No published figure exists for how many serviceable devices sit unused in storage. Estate: 1.05 devices per employee on a four-year refresh.
- What triggers it: A hardware request is raised while serviceable stock of the same class exists.
- Sources: no published source with a link. The price comes from retail listings.
Comatose server or VM decommissioned. One instance doing no useful work, switched off.
- Value of one unit (part published): An Asato estimate of the yearly cost of one instance: $1,800. The incidence is the published half: Koomey and Taylor (Anthesis, 2015 and 2017) reported 30% of servers as comatose, doing no useful work in six months. That data is old.
- How many (Asato estimate): 30 a year at 5,000 employees: 5% of instances, against the 30% that research reported. Estate: 120 instances per 1,000 staff.
- What triggers it: Sustained near-zero utilization with no dependent traffic.
- Sources: Koomey and Taylor, Zombie and comatose servers redux, Anthesis
Device recovered from a leaver. One laptop returned that would otherwise have been written off.
- Value of one unit (part published): The avoided replacement of a business-class laptop: $1,200 to $1,900 configured, from retail listings. The model uses $1,500. That the device would otherwise be written off is an Asato estimate.
- How many (Asato estimate): 18 a year per 1,000 employees, about one leaver in seven: an Asato estimate. For context, in Capterra's 2022 Employee Offboarding Survey of 287 respondents, 71% said at least one employee did not return company-owned equipment, worth $1,963 on average. Estate: 13% of staff leave in a year.
- What triggers it: The leaver date has passed, and the device is still assigned and has not checked in.
- Sources: Capterra, 2022 Employee Offboarding Survey; US Bureau of Labor Statistics, Job Openings and Labor Turnover, released March 13, 2026
Retired device sold for its residual value. One machine disposed of for value and not written off.
- Value of one unit (part published): A Forrester Total Economic Impact study commissioned by Apple (June 2026, 208 IT decision-makers) reports that PCs retain about 10% of their value after three years. 10% of a $1,500 laptop is $150. The model's refresh cycle is four years, so the real figure may be lower.
- How many (Asato estimate): 60 a year per 1,000 employees, roughly a quarter of the devices retired in a year: an Asato estimate. Not every retired machine has residual value worth chasing. Estate: a four-year refresh across the fleet.
- What triggers it: A device is marked for retirement, or its refresh date is reached.
- Sources: Forrester Total Economic Impact study, commissioned by Apple, June 2026
Rationalize your portfolio
Solution page: https://www.asato.ai/rationalize-portfolio
| Line | Value of one unit | Price basis | Units a year, before rounding to a whole unit | Estate it draws from | Share of estate |
|---|---|---|---|---|---|
| Duplicate application retired | $48,000 | Asato estimate | 3 x (employees / 5,000) to the power 0.35, x (software spend / its default) when that spend is below its default | No neutral count | Not stated |
| Never-deployed application canceled | $24,000 | Asato estimate | 2 x (employees / 5,000) to the power 0.35, x (software spend / its default) when that spend is below its default | No neutral count | Not stated |
Duplicate application retired. One whole application contract canceled where others do the same job.
- Value of one unit (Asato estimate): The value is the full contract: an Asato estimate of $48,000 a year.
- How many (Asato estimate): About three a year at 5,000 employees: an Asato estimate. For scale, Okta counts 101 apps per company on average among its customers, counting only apps connected through Okta, and a MuleSoft survey of 1,050 respondents reports 957 applications on average. Three a year is about 3% of the first count and under 1% of the second.
- What triggers it: A category holds four or more applications with overlapping function, and one has a low share of unique users.
- Sources: Okta, Businesses at Work 2025; Salesforce MuleSoft Connectivity Report 2026
Never-deployed application canceled. One application bought and never rolled out.
- Value of one unit (Asato estimate): An Asato estimate of a $24,000 contract. No published study separates applications never deployed from those under-deployed, so this is the softest line here.
- How many (Asato estimate): About two a year at 5,000 employees: an Asato estimate, set against the same two counts of applications: 101 per company among Okta's customers and 957 in the MuleSoft survey.
- What triggers it: Provisioned seats far below the purchased quantity for more than 180 days.
- Sources: Okta, Businesses at Work 2025; Salesforce MuleSoft Connectivity Report 2026
Stay ahead of contracts
Solution page: https://www.asato.ai/stay-ahead-of-contracts
| Line | Value of one unit | Price basis | Units a year, before rounding to a whole unit | Estate it draws from | Share of estate |
|---|---|---|---|---|---|
| Auto-renewal caught inside the notice window | $9,000 | Asato estimate | 12.5 x (employees / 5,000) to the power 0.35, x (software spend / its default) when that spend is below its default | No neutral count | Not stated |
| Third-party maintenance in place of OEM support | $4,200 | Part published | 7.5 x (employees / 5,000) to the power 0.7, x (hardware spend / its default) when that spend is below its default | 75 hardware support contracts at 5,000 employees | 10.0% |
| Cloud commitment coverage improved | $3,000 | Part published | 15 x (employees / 5,000) to the power 0.7, x (hardware spend / its default) when that spend is below its default | 600 server instances at 5,000 employees | 2.5% |
| Warranty renewed before expiry | $300 | Asato estimate | 25 x hardware spend / $390,000 | 260 warranties expiring a year per 1,000 employees | 9.6% |
Auto-renewal caught inside the notice window. One renewal reviewed before it rolls over, so the customer can renegotiate it. Asato prepares the brief, and the customer negotiates.
- Value of one unit (Asato estimate): An Asato estimate: about 15% of a $60,000 contract.
- How many (Asato estimate): 12.5 a year at 5,000 employees. No neutral count of renewals exists, so this count is an assumption. For context, Capterra's 2026 Software Buying Trends survey of more than 3,300 buyers reports that two-thirds experience disruptions during implementation, purchase regret, or both.
- What triggers it: The notice deadline for a renewal is within 30 days.
- Sources: Capterra, 2026 Software Buying Trends
Third-party maintenance in place of OEM support. One server, storage array or network device moved off OEM support.
- Value of one unit (part published): Gartner's 2019 market guide is reported to say that third-party maintenance contracts can offer 50% to 70% savings off net OEM support prices. The model applies the midpoint, 60%, to an Asato estimate of a $7,000 yearly OEM contract: $4,200.
- How many (Asato estimate): 7.5 a year at 5,000 employees: an Asato estimate. It depends heavily on how much owned infrastructure remains. Estate: 15 data center devices per 1,000 staff are still on OEM support.
- What triggers it: A support contract comes up for renewal on hardware past its OEM warranty.
- Sources: Gartner, Market Guide for Data Center and Network Third-Party Hardware Maintenance, August 29, 2019 (second-hand)
Cloud commitment coverage improved. One workload moved from on-demand pricing to a savings plan or reserved instance.
- Value of one unit (part published): AWS publishes discounts of up to 66% (Compute Savings Plans) and up to 72% (EC2 Instance Savings Plans) against on-demand prices, for commitments of one to three years. The model applies a conservative 50% to an Asato estimate of a $6,000 yearly on-demand workload: $3,000.
- How many (Asato estimate): 15 a year at 5,000 employees: an Asato estimate. For context, BCG wrote in July 2025 that studies indicate up to 30% of cloud spending is wasted, with quick wins of 6% to 14%. That describes the pool, not what this line recovers. Estate: 120 instances per 1,000 staff.
- What triggers it: Steady on-demand usage with no commitment coverage.
- Sources: AWS Savings Plans pricing; BCG, Cloud Cover, July 2025
Warranty renewed before expiry. One device kept under cover that would otherwise fall out of warranty.
- Value of one unit (Asato estimate): An Asato estimate of one avoided out-of-warranty repair: $300. No current published repair cost fits this segment.
- How many (Asato estimate): 25 a year per 1,000 employees, about a tenth of the warranties that expire in a year: an Asato estimate. The rest are close enough to replacement that extending cover is the wrong call. Estate: a three-year warranty against a four-year refresh.
- What triggers it: A warranty expires within 60 days on a device not scheduled for refresh.
- Sources: no published source with a link. The value is an Asato estimate.
Worked examples
With every spend field at its default and every line on:
| Employees | AI spend | Software spend | Hardware spend | Low | Expected | High | Expected per employee |
|---|---|---|---|---|---|---|---|
| 500 | $54,000 | $708,000 | $195,000 | $47,616 | $100,260 | $236,400 | $201 |
| 5,000 | $540,000 | $7,080,000 | $1,950,000 | $472,950 | $890,700 | $1,299,600 | $178 |
| 25,000 | $2,700,000 | $35,400,000 | $9,750,000 | $2,364,750 | $4,192,500 | $5,226,000 | $168 |
By solution at 5,000 employees:
| Solution | Low | Expected | High |
|---|---|---|---|
| Control AI spend | $54,000 | $76,950 | $76,950 |
| Right-size software | $418,950 | $426,150 | $488,550 |
| Stop paying for idle hardware | $0 | $309,000 | $309,000 |
| Rationalize your portfolio | $0 | $0 | $192,000 |
| Stay ahead of contracts | $0 | $78,600 | $233,100 |
Every line at 5,000 employees. Each value is the whole units times the value of one unit:
| Id | Line | Solution | Units a year | Value of one unit | Potential savings a year | Counts toward |
|---|---|---|---|---|---|---|
| a1 | Idle AI assistant seat reclaimed | Control AI spend | 150 | $360 | $54,000 | Low, expected and high |
| a3 | Overlapping AI assistant removed | Control AI spend | 75 | $228 | $17,100 | Expected and high |
| a2 | AI seat released on the leaver date | Control AI spend | 195 | $30 | $5,850 | Expected and high |
| e1 | Microsoft 365 E5 moved down to E3 | Right-size software | 175 | $252 | $44,100 | Low, expected and high |
| e2 | Microsoft 365 seat reclaimed | Right-size software | 200 | $468 | $93,600 | Low, expected and high |
| e3 | Adobe Creative Cloud reduced to a single app | Right-size software | 30 | $744 | $22,320 | Low, expected and high |
| e4 | Salesforce seat reclaimed | Right-size software | 40 | $2,340 | $93,600 | Low, expected and high |
| e5 | Salesforce edition moved down | Right-size software | 20 | $2,400 | $48,000 | Low, expected and high |
| e6 | Duplicate Power BI Pro license removed | Right-size software | 60 | $168 | $10,080 | Low, expected and high |
| e7 | Collaboration seat reclaimed | Right-size software | 275 | $390 | $107,250 | Low, expected and high |
| e8 | Docusign plan right-sized | Right-size software | 40 | $180 | $7,200 | Expected and high |
| e12 | Licenses released on the leaver date | Right-size software | 650 | $96 | $62,400 | High only |
| e14 | Orphaned account or OAuth grant revoked | Right-size software | 225 | Not priced | Not priced | Nothing: counted, not priced |
| e15 | Idle device redeployed, not bought | Stop paying for idle hardware | 50 | $1,500 | $75,000 | Expected and high |
| e19 | Comatose server or VM decommissioned | Stop paying for idle hardware | 30 | $1,800 | $54,000 | Expected and high |
| e13 | Device recovered from a leaver | Stop paying for idle hardware | 90 | $1,500 | $135,000 | Expected and high |
| e17 | Retired device sold for its residual value | Stop paying for idle hardware | 300 | $150 | $45,000 | Expected and high |
| e9 | Duplicate application retired | Rationalize your portfolio | 3 | $48,000 | $144,000 | High only |
| e10 | Never-deployed application canceled | Rationalize your portfolio | 2 | $24,000 | $48,000 | High only |
| e11 | Auto-renewal caught inside the notice window | Stay ahead of contracts | 13 | $9,000 | $117,000 | High only |
| e18 | Third-party maintenance in place of OEM support | Stay ahead of contracts | 8 | $4,200 | $33,600 | Expected and high |
| e20 | Cloud commitment coverage improved | Stay ahead of contracts | 15 | $3,000 | $45,000 | Expected and high |
| e16 | Warranty renewed before expiry | Stay ahead of contracts | 125 | $300 | $37,500 | High only |
With two spend fields overwritten (#n-12000-40-15): 12,000 employees, $40,000,000 of software spend (2.35 times the model's $16,992,000), $15,000,000 of hardware spend (3.21 times the model's $4,680,000), and AI spend at its default of $1,296,000. Low $2,496,456, expected $4,081,716, high $4,937,076, expected per employee $340.
| Solution | Low | Expected | High |
|---|---|---|---|
| Control AI spend | $129,600 | $184,680 | $184,680 |
| Right-size software | $2,366,856 | $2,407,536 | $2,557,296 |
| Stop paying for idle hardware | $0 | $1,346,700 | $1,346,700 |
| Rationalize your portfolio | $0 | $0 | $264,000 |
| Stay ahead of contracts | $0 | $142,800 | $584,400 |
The lines that follow the two overwritten fields, with the count at the default spend and the count after scaling:
| Line | Units at the model's default spend | Units a year | Value of one unit | Potential savings a year |
|---|---|---|---|---|
| Microsoft 365 E5 moved down to E3 | 420 | 989 | $252 | $249,228 |
| Microsoft 365 seat reclaimed | 480 | 1,130 | $468 | $528,840 |
| Adobe Creative Cloud reduced to a single app | 72 | 169 | $744 | $125,736 |
| Salesforce seat reclaimed | 96 | 226 | $2,340 | $528,840 |
| Salesforce edition moved down | 48 | 113 | $2,400 | $271,200 |
| Duplicate Power BI Pro license removed | 144 | 339 | $168 | $56,952 |
| Collaboration seat reclaimed | 660 | 1,554 | $390 | $606,060 |
| Docusign plan right-sized | 96 | 226 | $180 | $40,680 |
| Idle device redeployed, not bought | 120 | 385 | $1,500 | $577,500 |
| Retired device sold for its residual value | 720 | 2,308 | $150 | $346,200 |
| Warranty renewed before expiry | 300 | 962 | $300 | $288,600 |
With every spend field at zero (#n-5000-0-0-0): no line counts anything, and low, expected and high are each $0.
With hardware spend at half its default (#n-5000-d-0.975), a line that follows headcount and draws on hardware spend, "Device recovered from a leaver", reads:
18 a year per 1,000 employees is 90 at 5,000 employees. Your hardware spend is 0.5 times the model's $1,950,000 for that headcount, so the count scales to 45.
What the output is, and is not
- The output is an estimate of potential savings. It is not a measured result, and this page reports no customer results.
- The assumption that matters most is how often each event happens. Every frequency is an Asato estimate. The page has no field for an organization's own counts.
- The estate assumptions (for example, that 95% of staff hold a Microsoft 365 license) and the spend defaults are Asato estimates.
- The context figures quoted beside a line (surveys, government statistics, analyst statements) explain why a frequency is plausible. They are not the frequency.
- The portfolio lines and the renewal line have no neutral count to set their frequency against. Their units a year are plain Asato estimates.
- List prices are each vendor's own published prices. Microsoft 365 prices are those in effect from July 1, 2026.
- The page gives each line in about 60 words. The fuller text of each line is on this page, under "The saving lines, by solution".
Sources
Sources are vendor price pages, government statistics, analyst and research publications. A source marked second-hand was quoted from a secondary copy and has not been read at source.
- Microsoft 365 Copilot pricing
- GitHub Copilot plans
- Dillon and co-authors, NBER working paper 33795: a randomized trial of an AI assistant
- Ramp Economics Lab, Ramp AI Index, May 13, 2026
- a16z, AI in the enterprise, June 2025
- Microsoft 365 pricing and packaging update, effective July 1, 2026
- Microsoft Power BI pricing
- Adobe Creative Cloud for teams pricing
- Salesforce Sales Cloud pricing
- Slack pricing
- Box pricing
- Tableau pricing
- Figma pricing
- Docusign eSignature pricing
- US Government Accountability Office, GAO-24-105717, January 2024
- Gartner press release, July 2016, on cutting software costs (second-hand)
- US Bureau of Labor Statistics, Job Openings and Labor Turnover, released March 13, 2026
- Capterra, 2023 Insider Threats Survey
- Capterra, 2022 Employee Offboarding Survey
- Capterra, 2026 Software Buying Trends
- IBM and Ponemon, Cost of a Data Breach 2026
- Okta, Businesses at Work 2025
- Salesforce MuleSoft Connectivity Report 2026
- Forrester Total Economic Impact study, commissioned by Apple, June 2026
- Koomey and Taylor, Zombie and comatose servers redux, Anthesis
- Gartner, Market Guide for Data Center and Network Third-Party Hardware Maintenance, August 29, 2019 (second-hand)
- AWS Savings Plans pricing
- BCG, Cloud Cover, July 2025
Trust and governance
- The calculator runs in the page. It makes no network calls, and nothing a visitor types is stored or sent. A link to an estimate is built only when the visitor asks for one, and it goes to the visitor's clipboard.
- Asato does not negotiate prices. It equips the customer to negotiate.
- Nothing changes in a customer's systems until a named person approves it.
- The Asset Clarity Report is built with read-only access.
FAQs
Is the result a quote or a guarantee?
Neither. It is an estimate: the value of one unit multiplied by an estimated number of units, for an organization of the size and spend entered. The calculator has not seen the organization's estate. It is not a customer result and not a promise. The low end counts only lines priced from a vendor's published list price. The expected value adds lines where part of the value is published and part is an Asato estimate. The high end adds lines priced by Asato estimate alone. How often each event happens is an Asato estimate on every line, so even the low end is not a verified figure.
Why is there no line for AI billed by use?
No neutral rate exists. The AI lines cover assistant seats, which carry a published price. For AI billed by use, through a meter or an API, no neutral figure exists for how much could be recovered, so the calculator leaves it out and does not invent one.
Does Asato negotiate with vendors?
No. Asato equips the customer to negotiate: it supplies the count and the evidence behind it. The customer negotiates. Nothing changes until a named person approves it.
Is anything a visitor types stored or sent?
No. The calculator runs in the page. It makes no network calls, and nothing a visitor types is stored or sent. A link to an estimate is built only when the visitor asks for one, and it goes to the visitor's clipboard and nowhere else.
Next step
Get your Asset Clarity Report at https://www.asato.ai/talk-to-sales. It prices one area of the buyer's own estate in one week, with read-only access. If it finds no savings, the customer pays nothing.
Related pages
- Asato home
- Control AI spend
- Right-size software
- Stop paying for idle hardware
- Rationalize your portfolio
- Stay ahead of contracts
- Site index for agents
Last updated: 2026-10-09