Keep Score with Service Agreements

By Julie Gable

How do you know when you’ve done a good job? Unlike baseball, jobs in the information sector don’t produce batting averages or earned-run statistics. While objective measures are difficult to devise, they are highly necessary, particularly if you’re contemplating extensive outsourcing or the use of application service providers.

Wise companies use service level agreements, which spell out expectations and performance measures. The agreements detail a common understanding of excellence: not just what has to be delivered contractually, but how consistently and how well the job must be performed.

In some environments, achievement seems less important than the ability to consistently read minds; service level agreements, however, take the clairvoyance out of the customer relationship. The best service level agreements foster partnership, allowing providers to develop innovative solutions. The worst are like prenuptual agreements that kick in only when the relationship is beyond repair.

A service level agreement (SLA) should, of course, cover service — identifying what is provided and under what conditions, in what timeframes, to what standards and at what cost. The agreement should also cover management, describing how service effectiveness is tracked, how often report cards appear and how problems will be addressed and resolved. Most SLAs provide some remuneration if performance slips below agreed-upon levels. The more creative agreements recognize mutual benefit and reward providers when performance exceeds expectation.

Service level agreements are easier to create for mature industries without complex services, because both the customer and provider have experience to draw from. A typical agreement for document conversion services, for example, specifies prep standards, pickup and delivery schedules, quality standards for images, indexing accuracy and deadline for job completion. Typically compensation is awarded if standards or estimates are not met. It’s relatively straightforward: The service controls all the performance factors. Conversion service bureaus know from experience what they can reasonably handle.

The trend in using application service providers has presented a relatively new model for document imaging/management service level agreements. You rent storage, software, maintenance, back-up and security for your data and documents. You expect real-time access and fast retrieval, because data and documents are part of workflow processes. You also expect support, security and a safety net that guarantees that nothing is lost in the event of system failure, unforeseen disaster or similar crises.

Are your expectations reasonable? To some degree, the ASP’s origins — service bureau, software vendor, VAR or pure-play — influence its service levels.

Automated Document Management Solutions (www.docuclass.com), a conversion service that is an ASP for Hyland Onbase, operates an electronic repository. The company offers 24-hour standard technical support, but there are options for additional problem resolution. For example, a customer service software package called “Bean” provides real-time help that can walk users through retrieval while they are using the application. A monthly survey tool monitors customer satisfaction.

Another ASP, InterTech (www.intertech.com), uses special software to monitor and improve customer service. “Support has to go beyond the technical to resolve work process questions and problems,” says president Steve Hindman.

Even though ASP services are complex and providers don’t control all the performance elements, a service level agreement is a good idea. After all, you are paying for the service. In this embryonic industry, what can you reasonably expect? In creating an SLA, consider the following essentials.

Availability. Most ASPs will guarantee server availability (up and running) of 99.9% or better with 24-hour, seven-day-a-week access. One factor to consider is whether clusters of small servers are used or whether all customers share one large server. At what utilization levels will CPUs be upgraded?

System Performance. How much responsibility can the ASP take for performance? Throughput depends on bandwidth from the customer to the ISP and from the ASP to the customer. Users can achieve better performance with dedicated bandwidth, but it is expensive.

Application Performance. The main criteria is the number of users that can be supported, but ASP application performance also depends on other factors like bandwidth and the other applications running on the customer’s internal network.

Failover. How quickly does the backup system kick in when needed? What is the provision for data loss during system failure?

Clearly, developing service level agreements will be an iterative process as the application service provider industry develops. For now the trick is to strike a balance. Customers who micromanage every possible contingency won’t leave enough space for the provider to develop innovative solutions, and that will undercut the contributions that ASPs are poised to make.

— Julie Gable, CDIA, LIT, (juliegable@aol.com) is an independent consultant. Product or service mentioned should not be construed as an endorsement.