When On-Demand Isn't: Surviving a Hosted-CRM Failure

Updated: August 20, 2012

"I'm down (I'm really down)
I'm down (Down on the ground)
I'm down (I'm really down)
How can you laugh when you know I'm down?
How can you laugh when you know I'm down?"

Who knew Paul McCartney and John Lennon were so prescient in their understanding of how a CRM user would feel when his or her SaaS (software as a service) product failed him? "Down" in this context means that you can't use it. The provider may be functioning perfectly but if for any reason you can't reach the application — a network problem, a local outage or a bug in your own system — you're down.

In the worst case, an unavailable CRM solution means your sales force can't sell, your CSRs (customer support representative) can't help customers and the money-making part of your operation grinds to a screeching halt.

Of course, any kind of computer system — including the one in your office — can suffer an outage . SaaS vendors like to point out that with a remote system, you can continue to work even if a problem closes down your entire office. "Part of the point of [a remote] system is that if there's a problem in one place, you can go anywhere else," said Michael Fisher, CEO of ElephantDrive, a provider of backup and storage services. Service is available as long as users can get someplace that has connectivity.

ElephantDrive is familiar with outages from both sides. Not only does it offer remote backup and access services to its customers, but it is also a customer of Amazon.com's S3 hosted-storage service , which it uses to handle part of its storage. ElephantDrive is affected by Amazon's outages, just as its customers are affected by an outage at ElephantDrive.

Needless to say, ElephantDrive pays a lot of attention to level of service. "We monitor the system very closely and take [detailed] measures of the performance thousands of times during the day," Fisher said. "We look for degradation of service so we can add additional capacity. We look for early indicators of problems."

This kind of monitoring is actually fairly easy to do since both the computers and network components have logging built in. By feeding the reports of things like response time into the logging software and setting alarms to warn when parameters are moving significantly toward trouble, it's not difficult to get warning when something is going wrong.

Besides contacting your hosted-service provider or your ISP when you see a problem developing, you can also follow other protocols, like caching data locally, in preparation for a possible outage.

"We do things to minimize the impact," Fisher said. For example, ElephantDrive studies the way data is accessed by its users and saves frequently accessed data locally. By keeping copies of the data , the company can have it ready even if there is an interruption in service.

While hosted-CRM providers and VARs (value-added resellers) are eager to talk about features, they are much less interested in discussing service levels. That applies to the customer side of the equation as well. After all, that functionality is what customers are buying. They're less inclined to pay attention to the grubby back-end details like reliability.

The First Step Is Simply to Ask the Question

Start by asking, "What do we do if we lose service?" You'll start getting the answers. Some are immediately obvious. Others aren't.

Salesforce.com learned this the hard way . Most hosted CRM companies are still nervous about it.

Availability isn't just a job for the SaaS provider. The decisions that you and your company make will have a major influence on how susceptible you'll be to service interruptions. Although SaaS puts most of the hardware and administrative headaches on the service company, the configuration of your local systems is still going to have an important impact.

The Ideal Solution

The best situation, obviously, is not to have the problem. If your hosted-CRM solution never fails, then the problem doesn't arise. That's clearly impossible. What is possible, however, is making sure your chosen solution — from host software through communication channels to the applications on your salespeople's computers — is as robust and bulletproof as possible.

That includes redundancy, or having alternate pathways. Ideally, you should have a second connection to the Internet in case your main one goes down. The server on your end should have features like dual power supplies and a UPS (uninterruptible power supply) to keep it running in the event of problems. In general, you want to eliminate single points of failure wherever possible.

What's the Provider's Reputation?

Hosted-CRM reliability, like quality, is difficult to assess because it's made up of so many things. To be really reliable, a solution provider has to pay attention to dozens of factors, large and small.

Reputation is a touchstone for reliability. You should choose providers, both for a hosted solution and for your network services, based in large part on their reputation for reliability.

An Appropriate SLA Is Critical

Everyone in the CRM chain, from the software provider through the ISP and even your own equipment (if you have a maintenance contract), should be covered by SLAs (service-level agreements) specifying the amount of guaranteed availability, the steps that will be taken to ensure it and the appropriate penalties if the agreement isn't met.

SLAs can be pretty arcane and what you get often depends on negotiations. This is normally handled by your IT staff or your CRM consultant . Give them firm guidance and let them handle it.

Structure Your Solution for Reliability

In the context of your hosted-CRM solution, "solution" implies the entire package, application, provider, communications channel, host and everything else associated with it.

Design your system to be fault-tolerant. For instance, do you want to have your sales staff use a centrally stored up-to-the-minute schedule, or do you want them to work from a schedule which is automatically downloaded several times a day? The up-to-the minute schedule sounds wonderfully high tech, but what happens when service is interrupted? If the salespeople store their schedules on their own computers and simply have them updated automatically, they can continue to function uninterrupted even if the central computer isn't available.

In general, it's better to have as much of the action on your server or your sales staff's computers as possible.

You Need Backup Processes

In this case "backup processes" refers to the business procedures you use when your hosted applications are available — not backing up your computer data (although that's important too).

Backup process for CRM typically include some way to cache transactions and other information until the system is back. They should also include ways to transact basic business, such as order taking. An extreme example would be to use pencil and order pads for the duration, but you're probably not going to have to go that far.

Besides covering critical functions, your backup processes need to be standardized. In other words, things such as alternate order-entry procedures need to be spelled out in detail so that everyone does them the same way. This also needs to be integrated with departments such as accounting and shipping, which will be affected.

Test Everything Regularly and Often

All the disaster-recovery procedures in the world won't do you a bit of good if they don't work when you need them. Make sure you test your backup processes on a regular basis — say, once a quarter — to make sure everything works.