I gave a speech Monday in Oslo on Do It Wrong Quickly and, as usual, the best parts of the speech for me were the audience questions. One question in particular came up that I am asked frequently, so I thought you might like to hear the answer here. I speak frequently about how to transform your marketing and your customer Web sites based on experimentation and feedback metrics, but I am often asked if you can take the same approach on your intranet for your employees. My answer is, for the most part, yes!
That “for the most part” is a hedge, but overall I think it is the right approach to use with any intranet. Let’s look at how to do it.
Experimental marketing to customers is simpler in many ways than the same approach applied to employee intranets, in part because we know what the goal is of our marketing efforts—to sell stuff. We want to improve our Web site so that we sell more, but is that all there is to it? No.
The reason that it works is two-fold. Part of the secret is that we want to sell stuff but the other part is that our customers want to buy stuff. They are as motivated to make the purchase as we are to sell. They might have a problem to solve that requires the purchase, or maybe they just plain want it. (OMG, the new iPod just came out..)
Purchasing the product is what the customers are trying to do—it’s their task. We can apply many of the same “experiment with feedback” techniques to other tasks, whether they are performed by customers, employees, or someone else.
Let’s apply the same kind of thinking to technical support. It doesn’t matter whether you are a computer company helping customers with their laptops or you are an IT department providing a Help Desk for employees, the tasks are the same. How can you use experimentation and feedback to improve your Web site?
- Identify existing conversions. Find a set of tasks that have clear completion actions you can count, such as downloading a new driver for a laptop. If someone starts out searching for a new driver or navigates to the part of the site to download drivers, it’s clear what they are trying do. Count how often they succeed and then keep experimenting to improve that metric over time.
- Create new conversions. A lot of technical support questions don’t have clear success actions—it’s hard to tell whether people got the right answer to a problem just because they landed on an answer page—maybe it wasn’t the right answer and they just gave up. Create a conversion by surveying people as to whether the page was helpful and experiment to raise the results. Don’t worry about whether more negative people answer than positive ones—just focus on the trend from month to month being positive as you make changes.
- Use a negative conversion. Place a button on that same page that offers more help if the problem wasn’t solved (and focus on the pages that require the most extra help). Or track the number of people attempting to solve their problems on the Web versus those who call on the phone. See if you can lower the more expensive phone calls by solving more problems on the Web.
It’s not always easy to do these things—it’s often harder than tracking customer conversions from your experimental marketing—but it can be done. Not convinced? Let’s look at another example, the human resources part of your site.
You can identify many existing conversions for employee HR tasks, such as downloading the right form, signing up for employee health benefits, changing your 401(k)—you get the idea. There are myriad common tasks that can be tracked to completion. You can look at how many people search or navigate for these items and see how many succeed. Change it a bit and then see if the numbers are getting better.
You can also identify new conversions for some HR tasks. Think about the annual task of finding out which exact days are company holidays this year. Instead of just sticking them on a Web page that displays them, why not put a button on that page that automatically updates the employee’s personal calendar? When someone presses the button, you now know they got the answer.
And you can use the same kinds of negative conversions you did for tech support, by monitoring the number of e-mails and phone calls you receive for HR questions. Again it’s less important that these numbers be exact than that you observe the trends are going in the right direction.
Can you measure a conversion for every task your employees undertake on your intranet? Probably not. You’ll always have tasks that defy measurement. But if you apply the lessons you learn from the tasks you can measure to all your tasks, you have the best chance of improving your intranet globally.
It’s not easy to apply the “do it wrong quickly” approach to intranets, but I think it beats the alternative. Right now, what is a better way to improve intranets? I haven’t seen it.
