This post by Avinash Kaushik is reprinted with his generous permission. Originally published on his Occam’s Razor blog, it is the inspiration for this month’s issue of The Measurement Advisor, see our guide to the issue here. Avinash is our Measurement Maven of the Month, read more about him here.
If you bring sharp focus, you increase chances of attention being diverted to the right places. That in turn will drive smarter questions, which will elicit thoughtful answers from available data. The result will be data-influenced actions that result in a long-term strategic advantage.
It all starts with sharp focus.
Consider these three scenarios…
Your boss is waiting for you to present results on quarterly marketing performance, and you have 75 dense slides. In your heart you know this is crazy; she won’t understand a fraction of it. What do you do?
Your recent audit of the output of your analytics organization found that 160 analytics reports are delivered every month. You know this is way too many, way too often. How do you cull?
Your digital performance dashboard has 16 metrics along 9 dimensions, and you know that the font-size 6 text and sparkline sized charts make them incomprehensible. What’s the way forward?
If you find yourself in any of these scenarios, and your inner analysis ninja feels more like a reporting squirrel, it is ok. The first step is realizing that data is being used only to resolve the fear that not enough data is available. It’s not being selected strategically for the most meaningful and actionable insights.
As you accumulate more experience in your career, you’ll discover there are a cluster of simple strategies you can follow to pretty ruthlessly eliminate the riffraff and focus on the critical view. Here are are five that I tend to use a lot, they are easy to internalize, take sustained passion to execute, but always yield delightful results…
1. Focus only on KPIs, eliminate metrics.
Here are the definitions you’ll find in my books:
Metric: A metric is a number.
KPI: A key performance indicator (KPI) is a metric most closely tied to overall business success.
Time on Page is a metric. As is Impressions. So are Followers and Footsteps, Reach and Awareness, and Clicks and Gross Ratings Points.
Each hits the bar of being “interesting,” in a tactical oh that’s what’s happening in that silo soft of way. None, passes the simple closely tied to overall business success standard. In fact, hold on to your hats, a movement up or down 25% in any of those metrics may or may not have any impact on your core business outcomes.
Profit is obviously a KPI, as is Likelihood to Recommend. So too are Installs and Monthly Active Users, Orders and Loyalty, Assisted Conversions and Call Center Revenue.
Each KPI is of value in a strategic oh so that is why we are not making money or oh so that is why we had a fabulous quarter sort of way. A 25% movement in any of those KPIs could be the difference between everyone up and down getting a bonus or a part of the company facing layoffs. Often, even a 5% movement might be immensely material. What metric can say that?
When you find yourself experiencing data overload, don an assassin’s garb, identify the metrics and kill them. They are not tied to business success, and no senior leader will miss them. On the ground, people will use metrics as micro diagnostic instruments, but they already do that.
A sharp focus on KPIs requires concentrating on what matters most. Every business will have approximately six KPIs for a CEO. Those six will tie to another six supplied to the CMO.
After you go through the assassin’s garb process above, if it turns out that you have 28 KPIs… You need help. Hire a super-smart consultant immediately!
2. Focus only on KPIs that have pre-assigned targets.
This is a clever strategy, I think you are going to love it.
Targets are numerical values you have pre-determined as indicators success or failure.
Turns out, creating targets is insanely hard.
You have to be great at forecasting, competitive intelligence, investment planning, understanding past performance, organization changes and magic pixie dust (trust me on that one).
Hence, most companies will establish targets only for the KPIs deemed worthy of that hard work.
Guess what you should do with your time? Focus on analysis that is worth your hard work!
Start by looking at your slides/report/dashboard and identify the KPIs with established targets. Kill the rest.
Sure, there will be howls of protest. It’ll be John. Tell him that without targets you can’t identify if the performance is good or bad, a view every CEO deserves.
John will go away and do one of two things:
1. He will agree with you and focus on the KPIs that matter.
2. He will figure out how to get targets for all 32 metrics along all 18 dimensions.
You win either way. 🙂
An added benefit will be that with this sharp focus on targets, your company will get better at forecasting, competitive intelligence, investment planning, org changes, magic pixie dust and all the other things that over time become key assets. Oh, your Finance team will love you!
Special caution: Don’t ever forget your common sense, and strive for the Global Maxima. It is not uncommon for people to sandbag targets to ensure they earn a higher bonus. If your common sense suggests that the targets are far too low, show industry benchmarks. For example, the quarterly target may be 400,000 units sold. Common sense (and company love) tell you this seems low, so you check actuals to find that in the second month, units sold are already 380,000. Suspicion confirmed. You then check industry benchmarks: It is 1,800,000. WTH! In your CMO dashboard, report Actuals, Target and Benchmark. Let him or her reach an independent, more informed, conclusion about the company’s performance.
3. Focus on the outliers.
Turns out, you are the analyst for a multi-billion dollar corporation, with 98 truly justifiable KPIs (you are right: I’m struggling to breathe on hearing that justification, but let’s keep going). How do you focus on what matters most?
Focus your dashboards only on the KPIs where performance for that time period is three standard deviations away from the mean.
A small statistics detour.
If a data distribution is approximately normal then about 68 percent of the data values are within one standard deviation of the mean, about 95 percent are within two standard deviations, and about 99.7 percent lie within three standard deviations. [Wikipedia]
If your performance does not vary much, consider two standard deviations away from the mean. If the variation is quite significant, use six (only partly kidding!).
The point is, if performance is in the territory you expect, how important is it to tell our leaders: The performance is as it always is.
Look for the outliers, deeply analyze the causal factors that lead to them, and take that to the executives. They will give you a giant hug (and more importantly, a raise).
There are many ways to do approach this. Take this image from my January 2007 post: Analytics Tip #9: Leverage Statistical Control Limits…
Look for outliers. If you find them, dig deeper. If not, move on permanently, or at least for the current reporting cycle.
Use whichever statistical strategies you prefer to find your outliers. Focus sharply.
4. Cascade the analysis and responsibility for data.
In some instances you won’t be able to convince the senior leader to allow you to narrow your focus. He or she will still want tons of data, perhaps because you are new or you are still earning credibility. Maybe it is just who they are. Or they lack trust in their own organization. No problem.
Take the 32 metrics and KPIs that are going to the CMO. Pick six critical KPIs for the senior leader.
Cluster the remaining 26 metrics.
You’ll ask this question:
You might end up with eight for the VPs. Great.
Now ask this question:
You might end up with 14 for the directors.
Repeat it for managers, then marketers.
Typically, you’ll have none remaining for the Marketers.
Here’s your accomplishment: You’ve taken the 32 metrics that were being puked on the CMO and distributed them across the organization by level of responsibility. Furthermore, you’ve ensured everyone’s rowing in the same direction by creating a direct line of sight to the CMO’s six KPIs.
Pat yourself on the back. This is hard to do. Mom is proud!
Print the cascading map (CMO: 6 > VPs: 8 > Directors: 14 > Managers: 4), show it to the CMO to earn her or his confidence that you are not throwing away any data. You’ve simply ensured that each layer reporting to the CMO is focused on its most appropriate best sub-set, thus facilitating optimal accountability (and data snacking).
I’ll admit, this is hard to do.
You have to be deeply analytically savvy. You have to have acquired a rich understanding of the layers of the organization and what makes them tick. You have to be a persuasive communicator. And, be able to execute this in a way that demonstrates to the company that there’s real value in this cascade, that you are freeing up strategic thinking time.
You’ll recognize the overlap between the qualities I mention above and skills that drive fantastic data careers. That’s not a coincidence.
5. Get them hooked on text (out-of-sights).
If everything else fails, try this one. It is the hardest one because it’ll demand that you are truly an analysis ninja.
No senior executive wants data. It hurts me to write that, but it is true.
Every senior executive wants to be influenced by data and focus on solving problems that advance the business forward. The latter also happens to be their core competence, not the former.
Therefore, in the next iteration of the dashboard, add two more pieces of text for each metric:
1. Why did the metric perform this way?
Explain causal factors that influenced shifts. Basically, the out-of-sights (see TMAI #66 if you are a subscriber to my newsletter). Identifying the four attributes of an out-of-sight will require you to be an analysis ninja.
2. What actions should be taken?
Explain, based on causal factors, the recommended next step (or steps). This will require you to have deep relationships with the organization, and a solid understanding of its business strategy.
When you do this, you’ll begin to showcase multiple factors.
For the pointless metrics, neither the Why nor the What will have impact. The CMO will kill these in the first meeting.
For the decent metrics, it might take a meeting or three, but she’ll eventually acknowledge their lack of value and ask you to cascade them or kill them.
From those remaining, a handful will come to dominate the discussion, causing loads of arguments, and resulting in productive action. You’ll have known these are your KPIs, but it might take the CMO and her team a little while to get there.
After a few months, you’ll see that the data pukes have vanished. If you’ve done a really good job with the out-of-sights and actions, you’ll notice notice that the focus has shifted from the numbers to the text.
Massive. Yuge. Victory.
If more examples will be of value, I have two posts with illuminating examples that dive deeper into this strategy…
You don’t want to be a reporting squirrel, because over time, that job will sap your soul.
If you find yourself in that spot, try one of the strategies above. If you are desperate, try them all. Some will be easier in your situation, while others might be a bit harder. Regardless, if you give them a shot, you’ll turn the tide slowly. Even one month in, you’ll feel the warm glow in your heart that analysis ninjas feel all the time.
Oh, and your company will be data-influenced — and a lot more successful. Let’s consider that a nice side effect. 🙂
Knock ’em dead!
As always, it is your turn now.
Have you used any of the above mentioned strategies in your analytics practice? What other strategies have been effective in your company? What is the hardest metric to get rid of, and the hardest KPI to compute for your clients? Why do you think companies keep hanging on to 28 metric dashboards?
Please share your ideas, wild theories, practical tips and examples via comments.