SLAs Measure the Provider. XLAs Measure the Experience.
A service level agreement (SLA) measures whether IT, or an IT provider, did what it promised: systems were up 99.9% of the time, tickets were answered within four hours, incidents were resolved within the target window. Those are useful commitments. They say very little about whether employees could actually do their work.
An experience level agreement (XLA) measures the outcome from the employee's side. Instead of "the VPN was available," it asks "could remote employees connect quickly and stay connected, and how did they rate it?" XLAs combine technical experience data with employee sentiment, and tie both to a target the business cares about.
The watermelon effect. The term practitioners use for SLA reporting that is green on the outside and red on the inside: every metric is met, and employees are still frustrated. It happens because SLAs measure what is easy for the provider to count, not what employees experience. XLAs exist to close that gap.
What Makes an XLA
A workable XLA has five parts. Leave any of them out and it turns back into a dashboard.
| Component | What it is | Example |
|---|---|---|
| Scope | A specific employee journey or population, not "all of IT" | Remote and hybrid employees connecting and collaborating |
| Experience indicators | Technical measures taken at the endpoint, as employees experience them | Time to connect to VPN, share of poor-quality calls, application page-load time |
| Sentiment indicator | How employees rate the experience, collected in context | Short in-the-moment survey after calls, or a recurring pulse question |
| Target and baseline | Where you are today and where you commit to be | Baseline 6.1 out of 10; target 7.5 within two quarters |
| Owner and action | Who is accountable and what happens when the target is missed | DEX Lead owns it; a miss triggers a constraint review at the monthly program meeting |
How to Define Your First XLA
- Pick one journey that matters. Choose a population and workflow where friction is known and visible to leadership: onboarding new hires, remote collaboration, a frontline application, or Copilot and other AI tools. One well-run XLA beats ten that nobody acts on.
- Baseline both kinds of data. Measure the technical indicators for at least 30 days and run a sentiment collection on the same population over the same period. Without a baseline, a target is a guess.
- Choose two to four indicators, not twenty. Pick the measures that most directly explain the experience. If you cannot explain why an indicator moves the sentiment score, drop it.
- Set a target the business recognizes. Express it in terms leadership already uses: productive time returned, fewer escalations, higher satisfaction for a named group.
- Assign an owner and a response. Decide in advance what a miss triggers. An XLA with no consequence is a report.
- Review it in the governance cadence. XLAs belong in the monthly program review and the quarterly executive review described on our Governance page.
An Illustrative Example
The example below shows the shape of an XLA. The indicators and numbers are illustrative; real targets should come from your own baseline.
| XLA: Remote collaboration | Baseline | Target (2 quarters) |
|---|---|---|
| Share of Teams calls with poor audio or video quality | 9% | Under 4% |
| Median time to establish VPN connection | 38 seconds | Under 15 seconds |
| Employee rating of call experience (1 to 10) | 6.1 | 7.5 or higher |
| Owner | DEX Lead, with the network and collaboration teams; reviewed monthly | |
Common Pitfalls
- Too many indicators. A scorecard of 25 metrics hides the signal. Start with fewer and add only what explains the sentiment.
- Sentiment from the support sample only. Surveying employees after they open a ticket measures the minority who call. Collect sentiment from the whole population in scope.
- Targets set before the baseline. Committing to a number you have never measured invites either an easy win or a public miss.
- Penalties too early. With an outsourced provider, run XLAs alongside existing SLAs for a period before attaching commercial terms. Both sides need to trust the data first.
- Vendor-defined XLAs only. Platform scores are useful inputs, but the XLA should reflect your business priorities, not a default configuration.
XLAs With Managed Service Providers
Many enterprises meet XLAs first in an outsourcing contract. The practical path is to introduce them as "shadow" measures alongside existing SLAs: agree the scope, indicators, and data source, publish both sets of results for two or three quarters, and only then decide whether XLA results should influence commercial terms. That sequence gives the provider time to improve and gives you confidence the measures are fair.
Why XLAs Are a Leverage Point
Changing what an organization measures changes what it optimizes. That is why XLAs tend to produce more durable improvement than tool changes: they shift the goal of the system from "meet the provider's targets" to "deliver a good employee experience." Our Systems Thinking page explains why goals are among the highest leverage points in any system, following Donella Meadows.