Skip to content

ROI calculator

Model your own savings. Then check the working.

Five numbers about your organisation, and an indicative annual figure built only from published third-party research. Every rate is editable, and printed with the page that carries it.

  • Recalculates as you type
  • Every rate published by someone else
  • The arithmetic printed in full
  • Runs in your browser

Indicative annual saving

$1.26M

SaaS
$568,000
Cloud
$696,000
Three other classes
nothing

At the sample figures below. Change them and this follows.

Savings calculator

Savings model

Runs in your browser

Your figures

Your figures, not converted · the choice follows you

Everyone who holds a login, contractors included.

Desktops, laptops, servers, virtual machines, mobiles.

$

Everything billed by AWS, Azure, Google Cloud and the rest.

$

Subscriptions and seats, including the ones on expenses.

$

On-premises licences, maintenance and true-ups.

Five figures in · two rates out

Only cloud and SaaS spend meet a published rate. Software spend, devices and employees are context: they divide the total, and nothing multiplies them.

JavaScript is switched off, so the controls are hidden. Everything below is worked through at this sample organisation, at the published rates:

  • Employees 2,500
  • Managed devices 3,000
  • Annual cloud spend £2,400,000$2,400,000€2,400,000
  • Annual SaaS spend £1,600,000$1,600,000€1,600,000
  • Annual software & licence spend £2,000,000$2,000,000€2,000,000

Cloud + SaaS, at the two rates on record

$1.26M

Two published industry averages, applied to the figures you entered. No Certero rate goes into it; see what this model cannot know.

All five asset classes

spend, to scale

SaaS $1,600,000
35.5% of SaaS spend wasted on unused licences · $19.8M ÷ $55.7M $568,000
Cloud $2,400,000
29% of cloud spend is wasted $696,000
Software & licences $2,000,000
No published third-party rate · not modelled nothing
Devices 3,000 devices
No published third-party rate · not modelled nothing
AI no spend figure
No published third-party rate · not modelled nothing
  • Modelled saving
  • Spend with no published rate
  • No spend figure collected
Modelled saving, per employee, per year
$506
Annual spend the model has a rate for
$4,000,000
Cloud + SaaS spend per employee
$1,600
Licence spend per managed device
$667

Show the working

every number on this page, derived

  1. SaaS

    $1,600,000 annual SaaS spend
    35.5% of SaaS spend wasted on unused licences
    $568,000 modelled annual saving

    Derived: $19.8M wasted a year on unused SaaS licences ÷ $55.7M average annual SaaS spend. A separate published figure puts unused licences at 46%. That one counts seats, so the model divides the published waste by the published spend and uses what comes out.

  2. Cloud

    $2,400,000 annual cloud spend
    29% of cloud spend is wasted
    $696,000 modelled annual saving

    The published market figure, which went up this year for the first time in five.

  3. Software & licences

    Not modelled
    $2,000,000 annual software and licence spend
    no rate no published third-party rate, so nothing is modelled here
    — no modelled saving

    Nobody publishes a rate for this, so the lane stays empty. On-premises licence waste is metered title by title once the product is running, not estimated from an average.

  4. Devices

    Not modelled
    3,000 devices managed devices, as entered above
    no rate no published third-party rate, so nothing is modelled here
    — no modelled saving

    There is no published rate for hardware. What this lane returns is a device count, gathered by ten discovery methods on one inventory cycle.

  5. AI

    Not modelled
    not asked for we do not collect an AI spend figure
    no rate no published third-party rate, so nothing is modelled here
    — no modelled saving

    The published AI research measures growth rates only, so there is no waste rate for this lane to use.

  6. Total, indicative, per year

    The SaaS figure plus the cloud figure, rounded, because the model does not earn more precision than that.

    $1,264,000

The assumptions, and where they come from

There are two rates in the model, and both come from published research. The cloud rate is a published percentage. The SaaS rate is two published amounts divided by each other, with the division printed. Change either and everything above follows, and we will say so on the card.

SaaS spend wasted on unused licences

Derived: 35.5% $19.8M ÷ $55.7M

46% is the published licence-count figure. The model works from the two dollar amounts underneath it, and the rate that comes out is lower.

%

Two published amounts, one division between them, and no other step.

Cloud spend that is wasted

Published: 29%

The published figure, carried into the model unedited.

%

The research prints a single percentage, so that is what the box starts at.

Software & licences has no rate to edit, and neither do devices or AI: here is what we measure for those instead. Every link above opens the page that prints the figure beside it.

What the model rests on

Six published figures.
Two rates come out of them.

Every figure below is published third-party research. Only one of the two rates the model uses arrives as a rate; the other is divided out of two published amounts, and that division is set out below.

  • 305

    Context only

    SaaS applications in the average enterprise portfolio

    A count of applications, which sizes the problem the other three figures describe. Nothing in the model multiplies it. Portfolios held steady this year; the money inside them did not.

  • 46%

    A licence count

    of SaaS licences go unused — the average organisation uses 54%

    The model derives its SaaS rate from two dollar amounts instead of this one, and prints the division in the panel below.

  • +393%

    Context only

    growth in AI-native application spend at large enterprises

    Large enterprises specifically, and AI-native applications specifically. Across all organisations AI-native spend growth is +108%, and growth in use of applications across the whole AI category is +181%. The unqualified figure does not exist.

  • 29%

    Drives the model

    of cloud spend is wasted — up for the first time in five years

    One number for the whole market, published to the percentage point. This is the rate the cloud line multiplies, unchanged.

The SaaS rate, derived

two more published figures · one division

$19.8M wasted a year on unused SaaS licences
$55.7M average annual SaaS spend
35.5% of SaaS spend, the rate the SaaS line multiplies

46% is published as well, and it is the figure every vendor quotes. It counts licences, so multiplying your spend by it would price a dormant seat and an Oracle database the same way.

The application count and the AI growth figure describe the shape of the problem, and the model leaves both out of the arithmetic. Only the cloud figure arrives as a rate the model can multiply straight away.

Every figure links to the page that prints it, so you can check the arithmetic against the publisher. Both rates in the model are editable.

The empty rows

Three of the five,
we will not put a number on.

Nobody publishes a waste rate for hardware, on-premises licensing or AI spend, so the model leaves all three at zero. What follows is what the platform measures instead, and the shape of the answer that comes back.

01 / ITAM

Your managed devices

3,000 managed devices, at the figures you entered above

Modelled saving: zero

Hardware waste depends entirely on what is in your fleet: the age profile, the refresh cycle, how much of it is still on a support contract nobody uses. No industry percentage survives contact with that.

What the platform measures instead

The first number the platform produces is a count: the machines that respond to a sweep and appear in no record. Network Discovery covers a class-C subnet in under five seconds, then the native agent picks up Windows, macOS, Linux, AIX, HP-UX and Solaris on the same inventory cycle.

CerteroX ITAM
What comes back A count of machines
  • Agent
  • csinvcli
  • Agentless
  • Standalone
  • Network scan
  • Active Directory
  • Third-party ITAM import
  • Cloud & SaaS connectors
  • Browser monitoring
  • File metering

Ten discovery methods · one inventory cycle

answering on the subnet
in your CMDB today
the devices nobody knew about

The slots fill the first time discovery runs, one row per machine that answered.

02 / SAM

Your software and licences

$2,000,000 a year, and often the largest of the five

Modelled saving: zero

Licence waste is real. An Oracle environment and a Microsoft environment fail in completely different ways, and the variance between two organisations of the same size is enormous.

What the platform measures instead

Usage is metered from the files themselves, with first-used and last-used tracking and a rolling 90-day % Used figure per title. Above that sits an Effective Licence Position showing purchased, used, available, required, variance and exposure, across six publishers with dedicated engines.

CerteroX SAM
What comes back A licence position, per publisher
The columns of an Effective Licence Position. Values are filled as usage metering and entitlement records arrive.
publisher purchased used available required variance exposure
Microsoft
Oracle
IBM

Scroll for required, variance, exposure

% Used · rolling 90 days Three of six publishers with a dedicated engine

Microsoft, Oracle and IBM stand in for your own publishers here. The cells fill as usage metering and entitlement records arrive.

03 / AI

Your AI spend

+393% growth in AI-native application spend at large enterprises

Modelled saving: zero

AI spend is growing faster than anything else on this page, but growth is not waste, and the published research is all growth. There is no waste rate here for the model to use.

What the platform measures instead

Shadow AI detection classifies tools from their application feature tags, so the detection set grows on its own as new tools appear, and adoption is ranked by the share of your organisation using each tool. OAuth grants to AI tools are scored 0 to 100 on sensitivity, scope, consent and dormancy.

CerteroX AI Management
What comes back A list, and a score out of 100

OAuth grant risk · scored 0 to 100 on four dimensions

  • sensitivity
  • scope
  • consent
  • dormancy
0 100

3-tier adoption risk model Detection from application feature tags

Each bar carries a score once the connectors have read the grants your users have already given.

Not from the model

The only figures here
that were actually measured.

Everything above this line is two industry averages multiplied by numbers you typed. What follows is not. These are outcomes Certero publishes with the customer named and the case study behind them.

Every figure in these cards belongs to the organisation named on it.

“Certero’s SAM managed service allowed us to significantly mature our license posture at a fast pace, something that would have taken 3-4 years without their involvement.”
Reece Emson ITAM Asset/PSL Manager, NHS South West London ICB
£100k
Microsoft compliance risk mitigated
3–4 yrs
of SAM maturity accelerated
Read the case study
First look

Where the work starts.

The model produces two numbers. The larger one is where the work would start, and it names the product that would do that work.

On the figures above, the biggest modelled number is cloud waste, at $696,000 a year, which puts CerteroX Cloud Management first in the queue.

  • Cloud

    Twenty-six named checks against one billing account, among them Abandoned Kinesis Streams, Obsolete Snapshot Chains and Instances in Stopped State for a Long Time.

    Cloud Management
  • SaaS

    Unused licence detection at 30+ days of zero usage, then App Rationalization, which ranks overlapping applications by recoverable saving.

    SaaS Management
  • SAM

    An Effective Licence Position per publisher: purchased, used, available, required, variance, exposure.

    SAM
  • ITAM

    Network Discovery against a class-C subnet in under five seconds, and the gap it opens between a CMDB and what is actually answering.

    ITAM
  • AI

    The Shadow AI Dashboard: which AI tools are in use, by what share of the organisation, and which of them hold an OAuth grant, scored 0 to 100 on sensitivity, scope, consent and dormancy.

    AI Management

  • Three of the five

    SAM, ITAM and AI can never carry the badge above, because the model produces no figure for them. What those three return is a position, not a percentage.

    Why they are empty
Read this before you quote the number

What this model cannot know.

This is an indicative model built on published industry averages. Treat the figure it gives you as an estimate to argue with, and hold us to nothing we have not measured.

Four limits on the number
above.

  1. It has never seen your environment.

    It multiplies two industry averages by five figures you typed in. It knows nothing about your contracts, your commitment discounts, your renewal calendar, your reserved instance coverage, or the applications that are genuinely fully utilised at 100%.

  2. Identified waste is not money saved.

    The model estimates what published averages suggest is recoverable. It does not estimate how much of it you will act on, how quickly a licence can actually be harvested, what a renewal negotiation yields, or what it costs you to make the change.

  3. Averages hide enormous variance.

    A well-run FinOps practice will already be well below the published 29%. Somewhere nobody has looked in three years may be well above it. The average is a hypothesis about you, and the only way to settle it is to measure.

  4. Real savings depend on what you actually own.

    A number worth taking to a budget meeting has to come out of your own systems. That is what the platform does once it is running: metering usage title by title, costing the bill against twenty-six named checks, holding an Effective Licence Position that recomputes as entitlement lands. At that point the figures stop being averages.

About this calculator

The questions underneath the number.

The published licence-waste rate is 46%. Why does the model use 35.5%?

Because 46% is a share of licences and the model multiplies money. Applying a licence-count rate to a spend figure assumes every licence in a portfolio costs the same. The money is published as well: $19.8M wasted a year on unused licences, against $55.7M of average annual SaaS spend. Divide one by the other and you get 35.5% of spend, which is lower than 46% and is what the SaaS line uses. The division is printed in full under the research figures.

Why is the answer a single number rather than a range?

Because the figures underneath it are single numbers. Wasted cloud spend is published as 29%. The SaaS side is one waste amount and one spend amount, and one divided by the other is one number too. A range here would imply a precision the inputs do not have. The real uncertainty is whether two industry averages describe you at all, and widening the band would do nothing about that.

Why do three of the five asset classes produce nothing?

Because nobody publishes a waste rate for hardware fleets, on-premises licensing or AI spend. Those three sections set out what the platform measures instead, and what comes back.

Is there a Certero savings percentage in here anywhere?

No. Every rate in the model is a published industry figure, printed next to the number it produces. The only arithmetic that is ours is one division, and it is shown in full. Nothing in the total depends on a Certero figure.

Will Certero guarantee these savings?

No, and you should be wary of anyone who does. This model knows nothing about your contracts, your commitment discounts, your renewal calendar, your reserved instance coverage, or the applications that are genuinely fully utilised. It multiplies two industry averages by figures you typed in. It is a starting point for a conversation, not a business case.

What happens to the numbers I type in?

Nothing. The calculation runs entirely in your browser. There is no form, no submit button, no analytics event carrying your figures and no request to a server. Close the tab and it is gone.

Can I change the assumptions?

Yes, every rate in the model is editable and the page will follow. The moment a rate stops matching the published figure, the card it lives in is marked as edited, so you always know whether you are looking at the published research or at yourself.

Replace the averages

Now see which
mechanism produces it.

Bring the five figures you just typed and the assumptions you disagree with. We will walk the twenty-six cloud checks, the unused-licence detection and the Effective Licence Position at full scale, and show you which of them would move your figure and by how much.

Book a demo See the platform

No obligation, no gated PDF. If we are not the right fit, we will say so.