Garbage In…

As just noted, you can't measure what you are not capturing. If you don't capture the sources of your leads, you won't be able to compute the response rate on your marketing campaigns. If you don't log each prospect, you won't be able to calculate an accurate win/loss ratio. If you don't log each and every support case, you cannot hope to know your support volume. If you don't track the authors of knowledge base documents, you won't be able to reward the top contributors. Thus, the first step is to make sure that the data you want to measure is indeed available in the system. One obstacle may be the CRM tool itself. For instance, if the CRM system doesn't allow capturing the sources of your leads, you won't be able to track what promotions were most successful. If you want to calculate the effort time spent on each support case, you will find that most tools do not allow for easy tracking of effort time (at best, they let the reps enter effort time manually, which is neither user-friendly nor accurate). The major obstacle to data collection, however, is not a matter of any limitations in the tool itself but rather poor compliance by users in logging data and events. This can be because the users don't see the value of logging everything, because logging is overly cumbersome, because it's not enforced, or because counter-productive targets make it advantageous for the users to be selective in their logging efforts. The issue of counterproductive targets is particularly insidious. Let's say you decide to pay very careful attention to closing ratios and to give awards to reps with particularly high closing ratios. One of the likely consequences of such a system is that the more iffy deals will never get recorded in the first place, ensuring that close ratios remain very high. So you will get inflated, incorrect close ratios, but in addition, other analyses based on the data will be incorrect too. For instance, don't hope to perform any meaningful win/loss analysis based on a selected sample of deals. The cure here may be to ask the telesales reps to log the leads (assuming they are not measured on the close ratio). The same is true in the service realm. Say you have an initiative to decrease the number of interactions required to resolve service requests, correctly sensing that customers prefer to have their issues resolved in as few turns as possible. You run the very real risk of reps "forgetting" to log interactions on complex issues, impacting not only a proper count of interactions required to resolve issues, but undermining the entire case logging accuracy to boot. To avoid problems, use another measurement of quality in tandem, for instance implement a customer satisfaction survey. Be careful about setting goals and objectives that create incentives not to log events.

Get the Logging Religion

Beyond avoiding obvious disincentives to logging specific events, educate the users about the wisdom of logging all transactions not only because it's the expected standard of behavior, but also because it has benefits to them and to the organization.

  • It will help them keep track of where they are on each project, which is a great help in busy environments where users juggle many issues and many customers.
  • It will help their co-workers if they have to help with or take over a project, especially if there's no time for a debriefing. This should be a no-brainer in environments with a team approach.
  • It will help everyone figure out what happened if a win/loss or post-mortem analysis is required.
  • It will help the entire team by building up a knowledge base of best practices.
  • It will help the managers forecast resources and goals more accurately. Just think of what would happen if you forecast for a 2:3 close ratio when the real number is 2:10; or if you staff the support center for 30-minute cases but the average is really 45 minutes.

Logging compliance is an ongoing battle. Be prepared to monitor system usage and apply discipline accordingly if you want to ensure compliance.

Make it Easy to Log

While responsible end-users are quite aware of the benefits of logging all transactions, they still chafe at having to jump through hoops to record basic stuff or spending time recording exotic data that seems to never be used down the line. If you want good, reliable data, you need to make it easy for the users. Here are some examples.

  • Entering a new prospect should be done with one screen, with one click, and should allow for on-the-fly duplicate checking. Can your system do that? If not, why not?
  • Logging a simple support case (question and answer within a single phone interaction) shouldn't take more than two clicks: one to find the customer in the database, and one to close the case.
  • A recent (custom) system I worked with required support reps to enter no fewer than four categories to describe, respectively, the topic and sub-topic of the cause of the case, and the type and sub-subtype of the resolution of the case. That's too many! Simplify data entry to increase compliance.

Remember that the performance of the tool plays an important role in reinforcing the logging discipline. Slow tools automatically increase the likelihood of under-logging. If it takes an hour to download data to a laptop, sales reps will skip the synchronization and only use the system from the office, if at all. If the response time is bad in smaller offices, even desk-bound reps won't record all their activities. Make sure the system really works for everyone.

Capture Customer Satisfaction

As discussed earlier, a special concern for metrics is to capture direct measures of customer (or employee) satisfaction, which are not typically logged as a by-product of logging cases. I will discuss customer satisfaction here since it's the one most people focus on, but the same principles and strategies would apply to measure employee satisfaction. There are two ways to measure satisfaction, transactional surveys and overall surveys. Overall surveys are typically conducted once or twice a year, often include questions about many aspects of the relationship between the customer and the company, and gather feedback about many different interactions over a long period of time. A transactional survey, on the other hand, focuses on a particular interaction, be it a purchase or a support case, and is conducted soon after the fact, while memories are still fresh. Overall surveys are useful to get a periodic temperature of the customer base and also to survey all customers, even the ones that have not recently placed an order or a support request. Because of their scope, they tend to be quite long, creating a challenge to get customers to actually complete them. If you choose to do an overall survey, it's probably best to conduct it by phone to pretty much ensure that each customer who starts the survey finishes it. You may also want to give a small thank you gift to each customer who participates. Be careful with your sampling technique so you reach enough customers at minimum cost, and do keep the survey as short as you can: anything above 30 minutes is far too long, in my mind! Many specialized organizations offer outsourcing services for customer surveys if you prefer not to do it alone. While overall surveys have their place, my preference is for transactional surveys because they are specific, fast, and action-oriented. Because they focus on a specific transaction, transactional surveys allow you to trace both positive and negative feedback to a particular rep or a particular organization. In fact, you can use the satisfaction ratings to evaluate the performance of the specific individuals that helped the customer. Transactional surveys are fast. You get results the next day, not six or twelve months later. This is useful if you need to take action based on the feedback. Transactional surveys are action-oriented. They are not a mere measurement. You can use them to circle back to the customer with a resolution or thanks as needed. And you can do it fast since the results come back fast. The weakness of transactional surveys is that response rates can be low. What's even more worrisome is that customers who are particularly happy or particularly unhappy are more likely to respond than customers in the middle. This results in a biased sample. One of the best ways to combat it is to boost return rates. Here are three suggestions to do that.

  • Make it very easy to respond to the survey. In particular limit yourself to a handful of questions, say four or five (so we are talking a few minutes for a transactional survey versus a minimum of 15 to 30 on an overall survey). Use e-mail surveys for an unobtrusive delivery mechanism that allows customers to respond at their leisure. This is particularly effective to survey transactions that were conducted electronically since the customers are already comfortable with the medium.
  • Promptly get back to customers who are particularly happy or unhappy. Their likely response: "You guys actually read the surveys?!" Set thresholds for when to contact the customers and do so within 48 hours of the surveys for maximum impact.
  • Publish the results of the surveys to customers. This can take guts if the ratings are not that great. As a compromise, highlight specific changes you are making based on the surveys.

Finally, don't be overly obsessed by the potential for biased results. Do what you can to boost the response rates and then just work to improve the results, whatever they are. It doesn't matter if they are a bit biased.