End-user IT · Total cost of ownership
What does your IT estate actually cost?
Most enterprises budget end-user IT around the visible line items — devices, software, connectivity. The bigger numbers sit in support and downtime. Adjust the inputs to model your fleet. Totals and the breakdown update live. Defaults are illustrative — replace them with your real numbers before drawing conclusions.
Three buckets that overlap
The simplest mental model. Most cost lives in the overlaps, not the pure regions — that's where procurement teams under-count.
Hardware you own or lease
Per-seat software subscriptions
Consumption-based recurring spend
Devices
- PC / laptop
- Display
- Accessories
- Phone (handset)
Licenses
- Microsoft 365
- Adobe
- CRM seats
- Antivirus / MDM
Use cost
- Fixed data connectivity
- Mobile line / data
- AI inference / API usage
- Cloud storage / egress
Devices ∩ Licenses
- Surface + M365 bundle
- OEM Windows / Office
Devices ∩ Use cost
- Phone + mobile plan
- Laptop + tethered data
Licenses ∩ Use cost
- Copilot seat + tokens
- CRM seat + API calls
All three
- Mobile worker: phone + MDM + mobile data
Why the overlaps matter
A “phone” alone is a device. With a mobile plan it becomes device + use cost. With MDM and a Copilot seat layered on, it lands in all three buckets — which is why a single procurement decision (issue a phone) commits the company to ongoing spend across categories.
How to read this
The size of each region is illustrative, not to scale. The point is structural: most enterprise IT items sit in an overlap, and the recurring overlap spend almost always exceeds the one-time category spend.
14 drivers, ranked by typical share of TCO
Order is generalised across knowledge-worker enterprises, based on the Gartner end-user TCO framework and Forrester TEI methodology. Click any row for the reasoning. Bars show relative weight, not exact percentages — your fleet will differ.
The iceberg — visible vs hidden cost
If your TCO model only sums vendor invoices, you are measuring the tip. The water line is where most TCO arguments go wrong.
VISIBLE: Devices, SaaS / licenses, Connectivity, Security tooling, AI / inference, Accessories.
HIDDEN: IT support & service desk, Lifecycle ops (image, patch, refresh), End-user productivity loss, Self-support & “futz” time, Onboarding / training, Asset / license management, Logistics & shipping, Energy, Loss / theft / damage.
Why the split is so uneven
Visible costs are bounded by contracts. Hidden costs scale with how well (or badly) the visible stack is engineered. A cheap laptop with a slow image generates more tickets, more downtime, and more peer-support time — none of which appears on the laptop's purchase order.
The Gartner finding
In Gartner's classic end-user computing TCO model (originally developed by Bill Kirwin), “end-user operations” — downtime, self-support, peer support, casual learning — typically accounts for the largest single share of true TCO, often around half. IT operations is the next largest. Capital costs are usually the smallest of the three.
The implication: optimising only what's measurable on a PO will systematically rank levers wrong. Cutting device cost by 20% looks like a win on the invoice and can be a loss on TCO if it adds even a few hours of annual downtime per user.
How levers move costs across categories
Pulling one lever rarely affects only one category. This matrix maps common decisions onto their directional impact on each cost bucket — including the tradeoffs that get glossed over in single-line business cases.
| Lever | Devices | Software | Connectivity | Security | IT support | Productivity loss |
|---|---|---|---|---|---|---|
| Extend refresh cycle 3 → 5 yrs | ↓↓ | · | · | ↑ | ↑ | ↑↑ |
| Standardise on premium devices | ↑↑ | · | · | ↓ | ↓↓ | ↓↓ |
| Consolidate SaaS portfolio | · | ↓↓ | · | ↓ | ↓ | ↓ |
| Roll out AI assistants (Copilot etc.) | · | ↑↑ | ↑ | ↑ | · | ↓↓ |
| Adopt zero-trust / strong endpoint security | · | ↑ | · | ↑↑ | ↑ | ↑ |
| Move to remote-first / hybrid | ↑ | · | ↑↑ | ↑ | ↑ | · |
| Outsource L1/L2 helpdesk | · | · | · | · | ↓ | ↑ |
| Self-service portal + automation | · | ↑ | · | · | ↓↓ | ↓ |
The most common procurement mistake is reading only one column. Extending the refresh cycle is almost always pitched on the Devices column alone — but the Productivity-loss and IT-support columns usually swallow the savings on a fleet older than ~4 years.
Atlas
What to actually do with this
Six practical takeaways for IT and procurement teams running a TCO review.
1. Instrument productivity loss before debating refresh cycle
You cannot trade off device cost vs. downtime if you only measure one. Even a rough proxy — average ticket count per user × average resolution time × loaded hourly cost — moves the conversation from “the laptop is too expensive” to “the laptop saved 11 hours per user per year.” Without it, every TCO debate defaults to the invoice.
2. Treat SaaS and AI as one growing line item
Per-seat SaaS has roughly doubled across the last decade. AI tools layer on top with a different pricing shape (seat + tokens). For most knowledge-worker organisations, software now rivals or exceeds amortised hardware spend per seat — and unlike hardware, it doesn't get cheaper with age.
3. Audit the overlap regions
The overlaps (phone+plan, device+OEM-license, AI seat+tokens) are where shadow growth hides. They get budgeted in one category but renew in another, so neither owner sees the total.
4. Read the matrix, not the column
Single-column business cases for cost cuts (refresh, outsourcing, security cuts) are the most common source of TCO regret. Force every cost-reduction proposal to fill in the full row of the interaction matrix.
5. Don't optimise the cheap parts
Accessories, energy, logistics and insurance combined are usually less than 10% of TCO. The temptation is to negotiate them because they're easy. Time spent there has roughly 1/10th the leverage of time spent on support automation or device standardisation.
6. Run the calculator with three scenarios
Best fleet you've ever seen, worst fleet you've ever seen, and your fleet today. The spread is what you're really managing — and it tells you which lever has the most headroom in your environment specifically.
Sources: Gartner — End-User Computing TCO model (Bill Kirwin et al.); Forrester — Total Economic Impact methodology; Microsoft public list pricing for M365 and Copilot; Spiceworks “State of IT” reports for IT-to-user ratio benchmarks. All calculator defaults are illustrative and editable; nothing on this page claims a precise TCO figure for your organisation.
Method: Gartner end-user computing TCO framework (Bill Kirwin et al.) and Forrester Total Economic Impact methodology. Hidden costs — IT support labour and productivity loss — typically account for around half of true TCO. Nothing here claims a precise figure for your organisation; replace the defaults with your own data.