- Enterprises should work with providers to develop more meaningful KPIs within SLA agreements.
- AI technology is enabling predictive fault detection; working with providers will help improve AI performance.
Service level agreements (SLAs) are one of the tech world’s necessary evils and often one of its most ineffective tools. Ideally, no SLAs should be required; enterprises would much rather have the service work than be compensated for its failure. Furthermore, the financial penalty in SLAs is often nowhere close to compensating for the financial loss caused by an outage. GlobalData’s conversations with enterprises have revealed a lack of faith from enterprises in SLAs, with more than one IT manager referring to them as ‘pointless.’
The concept of an SLA seems fair enough; things sometimes go wrong, and when they do, the client will be compensated. However, it is also possible to view SLAs as an expectation for – and acceptance of – failure. Key performance indicators (KPIs) within SLAs also add to a feeling of alienation from businesses. Technical language about performance metrics (e.g., jitter, latency, and service uptime) matches the way that service providers think about networks and session initiation protocol trunks, but the language does not necessarily tally with what an enterprise needs from its service. As service providers seek to grow their managed and professional services revenues, enterprises have every right to expect the guarantees that come with a service to match their ambitions more closely.
There are two key areas in which change is, slowly, happening. The first is centered around ‘outcome-oriented’ SLAs. This essentially means that the KPI in an SLA will be tailored to match the outcomes that businesses wish to achieve from technology solutions rather than being purely centered on technical performance.
Virgin Media O2’s (VMO2) business-to-business unit has made a foray into this market with its new ‘Success Agreements.’ VMO2 has added increased flexibility to its SLAs with a greater tolerance for service churn within contracts and three months of free parallel running during provider transition periods. The success component of the SLA involves agreed-upon KPIs between the customer and the provider that will be monitored by a ‘success board.’ If success standards are not met, this may ultimately lead to zero charging for service components of contracts during periods where agreed levels of success are not achieved.
VMO2’s approach doesn’t drastically move SLAs away from traditional telecoms performance metrics, but it gives the customer greater ability to determine what is important and shows a willingness from the provider to take on more risk. The longer-term challenge will be finding a balance that shares risk and reward (on both sides) without the potential of either increasing exponentially.
Another aspect of the evolution of SLAs and wider service monitoring is being driven by artificial intelligence (AI) in combination with process automation. Orange Business Services has recently delivered an example of how AI can improve service performance with its new Service Manage-Watch proposition (for a full analysis, please see: “Orange Adds Predictive AI Power to its Network Monitoring with Service Manage-Watch Launch“). Orange’s solution commits to 10% of fault tickets being generated automatically by its AI engine. This is a tentative step, one that does not fully address the failings of current SLAs, but it reduces the burden on enterprises of reporting faults; it also includes the ability to diagnose errors more accurately with recommendations of how to prevent similar errors from occurring.
Both of these examples require cooperation between the enterprise and the provider, and it is clear that greater pre and post-sales engagement will increasingly deliver an enhanced SLA experience.