27 May

Do these metrics make me look fat?

scale

After I had my third kid I discovered that I was fat.

Not in a “do these jeans make my butt look big” way, but in an objective, measurable way (because that’s what I’m all about).  My BMI (Body Mass Index) was officially above “normal” and into “overweight”.

Being a researcher, I looked into various strategies for shedding those pounds and landed on a system that centered on counting “points”.  You were allowed to eat a certain number of points a day. Every food had a certain amount of points associated with them and when you exercised, you earned points for the effort.

I was measuring points in and out and my metric of success was the number on the scale every week.

But here is a truism about any kind of data collection attached to behavior:  You will optimize for whatever you are measuring.

I discovered that I was not choosing the right metrics.  As a result, I was nudging myself to behave in a way that was not optimal for my health.

I was choosing processed food over natural food because they had lower points.  Also, I avoided lifting heavy weights because that made the scale move up, not down.  I was indeed losing weight and meeting my metrics, but I wasn’t getting healthier.   By switching my metrics to better ones like body fat percentage and gym performance, I found that my behavior naturally followed to be more healthy.  I still lost the weight, and I totally rock those jeans now.

If you are using web analytics tools out of the box, you are probably making this same mistake, and the health of your organization is suffering as a result.

Most people use page views (number of visitors) and bounce rates (the percentage of people who leave after a very short time) as the fundamentals of their web metrics.

More  page views = good.

High Bounce rate = bad.

At first glance, this seems reasonable.  If those numbers are good, it means that lots of people are coming to your site and are finding a reason to stay.

But stop and think a minute about whether or not these metrics are the best ones for you.

In one my recent projects, the web department of a large company diligently reported on page views and bounce rates.  Many of their limited resources were were focused on increasing the number of visitors to the site.

Here’s the problem with that.  A lot of their page views were from locations that they did not serve.  That meant that many of their visitors had no possibility of ever becoming customers.  The website visitors that they were attracting from Australia didn’t contribute to their business at all.

At the same time, they were missing much more important metrics.  For example, how successful were people who wanted to pay their bills online?   Appallingly unsuccessful as it turns out.  They had only a 2% conversion rate.  That means thatout of every 1000 people who tried to give them money, only 20 were able to do so.   Getting revenue affects their bottom line much more than page views. However, the team was not reporting that number, so they were not doing anything to improve it.

Let me say that again in case you missed it.  980 people out of 1000 who tried to give them money, were unsuccessful.  They had to call tech support for help and wound up either paying over the phone, or mailing in a check.  Think about what that inefficiency cost their business!  Not to mention creating customer frustration.  And yet, somehow that number didn’t seem as important as page views.  Because of improper metrics, they had lots of people working on the “how do we get more visitors?” problem.  But no one working on the “how can we let people to give us money?” problem.  Pop Quiz: Which problem do you think was more important to their business?

Bounce rates can also be deceptive.  While in general it is true that it would be nice if visitors to your site found your content “engaging” and hung out and read it over a coffee, this is not always the best metric.  My client had several transactional processes such as registering workshops with an average time on the page of 4 minutes.  This looks like a good bounce rate at first glance but it is actually bad news.  If your visitors spend 4 minutes on the registration landing page, it means they are confused about the next step, not engaged.  Confused customers are bad.  But if you are not paying attention to the right metric, you won’t notice.

Here is the key point.  When you set up your Google Analytics, make sure that you think carefully about which metrics will mean a healthier business for you.  Most of the time, it won’t be page views.  Yes, you will have to do some customization of your analytics, but then you can be content in the knowledge that you are measuring the things that will push you in the direction of making your organization healthier and more profitable.

If you find yourself caring more about web site visitors who will never be customers, than about letting your actual customers give you money, that is a sign you need to rethink your metrics.

photo credit: Alan Cleaver (creative commons commercial license)

If you enjoyed this article, sign up for my newsletter!  Fresh stuff every two weeks about how listening to your customers can improve your bottom line.

13 May

Social Media Prediction: Lessons from Billboard’s Top 40 chart

fortuneTeller_smaller

What if you had a way to predict the future?

It turns out that by doing a relatively straightforward interpretation of social media conversations, you can make some educated guesses about what people are going to be discussing the next day.

If you are monitoring what people are saying about your brand on social media (and you should be), the ability to predict what people will be saying can help you address a crisis before it becomes a crisis.

If you are a political party or candidate that is following what your citizens are talking about on social media (and you should be), this can give you the ability to know what issues are going to be hot tomorrow.  When you know that, you can get out in front of issues just before they are about to hit big.

To understand how you might do this, I want you to stop thinking about social media for a  second and think about the Billboard Top 40 list of pop music.

In the Top 40 list, the number 1 song is vastly more popular than the number 40 song.  But once a song enters the top 40 list, you want to keep an eye on it – it might be the next mega-hit.  If you were to place bets on which song will make it to the Top 10 list next week, your odds would be better on song #38 than on song #88.

You see a similar pattern in the words that get used in social media conversations.

If you do any search on the Twitter stream on any given day, there will be a handful of words that are mentioned the most frequently.   Let’s call these the “Top Words”.  This is analogous to the Billboard Top 10 list.   The top words tell you what is dominating the conversation on that day.

But what if you want to know what will be dominating the conversation tomorrow,  not today?

For that, you can look at the “Secondary Words”.   Think of these as the songs that are numbers 11 through 40 on the Billboard list.

Secondary words are words that are:

  • On their way out of being Top Words
  • On their way towards being Top Words
  • Will sit in that category for a while and then move out, never making it to the top

Imagine that for every word that appears in your search, you create a bar with the height equal to how many times that word was mentioned.  You end up with a graph that looks like this.

TopWords

No matter what the search is, and what the day is, the graph always ends up having this kind of shape.  Usually, there are only a few words that appear very frequently (Top Words), some more words that appear somewhat less frequently (Secondary Words), and then most of the words don’t occur especially frequently.

Just like songs that are number 11 through 40, the secondary words, can give you very strong clues as to what is going to be a big topic the next day.

Imagine that you are a brand that has just had some kind of PR slip up.  You can do this kind of analysis to determine how worried you should be about it.  Are the words related to the scandal in the long tail of words?  If so, you can probably rest easy.  If the scandal words are in the secondary category you can start bracing yourself with your PR machine.

That’s it.  That’s the trick to predicting the future.

This kind of analysis is so straightforward, I am surprised that it is not supported by more social media analytics tools.  I have evaluated a lot of them and Nexamaster by Nexalogy is the only tool that I know of that makes this kind of interpretation a cinch without having to export your data to Excel and do it manually.

Here is the key point. The conversations that people are having about you on social media tell you what they are thinking about.   Listen effectively, and you will know what they might be thinking about tomorrow.