Whether you are setting up a new contact center environment or updating an existing one, there are many factors to take into account. One wrong move and when you go live, the whole structure can come tumbling down. That’s why thorough pre-deployment testing is crucial.
Once you have completed the steps to prepare for testing your contact center deployment (see Top 10 Tips for Testing Your Contact Center Network), it’s a good time to discuss the next phase of your journey: successful test execution. There are a number of subjects to consider when evaluating how your test will be completed.
Think long term
Often at Empirix we work with customers who don’t fully appreciate that, realistically, at some point (perhaps in the not too distant future) they will outgrow their current testing solution. They focus only on the here and now. Then, as time goes on and they continue to grow, they are required to change their preferred test deployment strategy to another model. This is not only disruptive, it’s also expensive and time-consuming.
You must understand both where you are currently and where you are going with your contact center when considering your testing plan. To achieve that goal, you should think in a three-year horizon about a few categories:
Growth of contact center interactions
- Does your company expect to acquire new customers over the three year horizon? Will they be calling the contact center?
- Will the growth be through internal acquisition programs?
- Is your company planning to make any acquisitions, thereby driving possible consolidation and interoperability of contact centers?
- Do you have a roadmap of new and replacement equipment that will be deployed in your contact center?
- Are you moving to a SIP-based platform?
- Are you changing vendors?
- Do you know of all your call flows? Are they documented?
- What is does the software upgrade roadmap look like? And at what frequency will upgrades occur?
- What effect will WebRTC have on the contact center?
Mix of multichannel contact
- Is your contact center primarily voice today or multichannel? What is the mix of voice, video and web interaction?
- Are you planning for video to be introduced?
- How might the mix of channels affect the quality of any single or combination of channels?
Whatever deployment method you choose (see below), be sure it aligns to the employee model your company has chosen.
- Are you outsourcing?
- What is your retention rate of employees with the skill to conducting testing?
- Are you confident in the expertise of your internal employees who will conduct testing?
- Might you need to bring in outside experts to assist in certain testing situations?
Consider your deployment model options
Choose a strategy that fits your goals and is consistent with your three-year horizon.
In this case, hardware and software are delivered to your location – you license them – and, generally speaking, all call scripting and testing is completed by you and your team. Some companies prefer to license and conduct testing themselves because there can be a reasonable amount of flexibility and short term cost-saving in this approach. The biggest problem is usually around maintaining knowledge and expertise of the nuances of testing.
Pure managed service
In this model, you coordinate with your testing provider who will conduct all testing remotely and then issue you a report. You do not have to worry about licensing equipment and software, but there is additional coordination and scheduling with your provider as well as a certain amount of costs around hiring an outside vendor.
Hybrid of previous models
In this model, you can license software and hardware for discrete tests. However, to conduct true customer experience testing at realistic volume, you would likely work with providers who maintain expertise and have remote access to on-premise test equipment. Many companies consider this a “best of both worlds” situation.
Test realistic scenarios
Make sure you can generate a large volume of interactions and can do so locally.
Companies usually don’t discover problems such as bottlenecks or outages at unrealistically low call volumes. It’s at high traffic times when your system is slammed that issues crop up – and when your customers call in for a great sale or require assistance during an emergency, you don’t want your system to fail. Testing at peak load conditions truly gives you a realistic view of your customers’ experience. The two key types of tests to conduct at load are:
- Regression testing. You must perform regression testing to actively test call paths and routing. While regression testing is important in all test engagements, it is super critical when changing vendors on whose equipment your auto attendant and call manager function. Often times call flows and routing tables are not well maintained and must be both documented before the new equipment or software is installed, and validated once the new installation is complete.
- Feature testing. You must also conduct feature testing to insure that all the features of your new systems are working.
Countries or locations
In this step, think about what makes sense based on your customer profile: Where will most of the calls be coming from? Where will they be going? Once you have answered those questions, generate test traffic from the appropriate countries or locations. Why do this step? Well, there are a few reasons. First, you don’t want to pay for toll charges, which can be brutal. This issue might occur, for example, when you are dealing with a heavy volume of calls that originating in the US and are delivered in a different country. Second, you should test as realistically as possible to insure that the quality of the customer experience is transparent and accurate. Make sure your provider has points-of-presences (POPs) worldwide to generate the volume of traffic required, and also as local as possible to provide a great view in to the customer experience.
Each of these three tips should be considered carefully as you map out your execution plan to conduct testing of your contact center. Do you have any other pre-deployment testing considerations you would add to this list? Let me know in the comments below.