As you go about your business of absorbing data, do you sometimes look at what’s being presented and let out a yelp of despair? Maybe 18 times a day?

No? Just me? :)

The issues that spark this are not huge things like fundamental problems with methodology or a lack of imaginative thinking. They are often little things that should be caught—if someone, somewhere along the way applied the wisdom they’ve accumulated over time.

Here are four commonly occurring mistakes, among many, that break my heart… Each is sourced from considering a data consumer’s perspective (probably your boss’s boss).

1. Calibrate altitude optimally, please!

This is such a simple mistake. Look—when you report numbers, ensure that they are at the same altitude.

Take this example from YouTube:

Seeing 11k Up and 351 Down gives 351 too much weight.

Subliminal, but super important in world where we are trying to quickly assess a lot of data.

An optimally calibrated version would read: 11k | 0.4k.
Or: 11,000 | 351.

Either strategy puts 351 in perspective. It is a much smaller number than 11,000.

Here’s an example from Twitter Analytics:

14.6k does not look as inefficient as it should when you see 18 and 12.
Fixing altitude you see: 14,600 | 18 | 12.

The value you are receiving for your money is now staring you in the face a bit more starkly.

That’s your job. Put things in context and sharpen reality.

[Before I forget, there’s a request for your help at the end of this email. Merci!]

2. Eliminate false precision.

On a base of forty-six million, one hundred sixty-five thousand, six hundred and eighty-eight dollars... Does the addition of eighty cents really add value?

Are you honestly that precise in your measurement?

The answer, across all modes of analytics, not just Google Analytics, is an emphatic heck no!

So, why imply that the numbers are more accurate than they really are? Why add to the visual clutter?

Ponder that box just under Product Revenue. Read the whole thing—consider what it is saying, and the value of saying it. Heartbreaking, right?

There are examples of false precision everywhere. I call them false because if you simply understand how the data is collected, processed and reported, you'll realize that the number could easily be plus or minus 5 to 10%. Yet, our presentation of them implies precision to a two-decimal-point level—which does not exist.

Take this table as an example...

You would lose none of the value for Bounce Rate by simply rounding the numbers up. 85.02% can just be 85%. 28.97% can be 29%.

When your leaders ask for two decimal points, likely to happen when the source of your pork products fly, explain why that would be a case of false precision. Amazingly, the explanation would cause them to trust you more.

When you report the results of your statistical analysis, share your confidence intervals. Say that you are saying 35%, but the confidence intervals stretch 4% to 43%, thus eliminating false precision.
When you report the results of your surveys, report that n=314 and most respondents were from the state of Illinois. Expose false precision.

So on, and so forth.

3. A pinch of context makes data delicious.

Don't scroll beyond the picture below, trust me.

Look at this slide, consider the number, and reflect on what you learned from it.

Go back.

Please. Ponder that question: what did you learn from the slide that was of value?

Ready?

On the surface, I'm sure you thought it was amazing. 80% is a huge jump in just two years.

Upon a bit more reflection, I suspect you thought… Wait, is it really big? Is it important? I feel empty inside.

That feeling is because the slide has no context.

If all other mobile searches grew by 6,000%, then 80% is terrible. 6,000% is context.

If 80% reflects only Android phones in Canada, maybe that is OK? Android and Canada are bits of context.

If the 80% increase in "best" mobile searches represents super-rich people, that is attractive (money!). That piece of psychographic information is context.

If the 80% increase resulted in an additional 76% increase in sales of mobile phones via mobile searches, omg that is amazing. That sales number is context.

If that 80% increase in mobile searches was during a time when desktop searches declined by 50%, that is scary (and important). That desktop searches number is context.

I could keep going, but you get the idea.

The reason an initially attractive piece of data made you feel empty was that it was not surrounded with context.

Any piece of data can be good or bad, useful or trash, insightful or a distraction. What puts it in one category or the other is context.

Most metrics are still reported without context—thus effectively neutering any value they carry.

You worked very, very hard on creating that report. Don't dump it on your leadership teams without ensuring that it has the most relevant bit of context attached to it.

If you want to play the game again, here's an example you can use... Email me back with the most amazing bit of context you would attach to this slide:

I can think of 16. You have to think of just one. Email me.

Here's a hidden tip for an analyst seeking relevant context: You'll need to understand business priorities, goals and strategy. That, all by itself, has the power to increase your influence—beyond making your reports and slides more valuable.

Context is queen. (Five still-relevant tips from, I kid you not, 2008!)

4. Don't throw the kitchen sink at your readers!

This is a rookie mistake I see everywhere.

Reports that have 16 columns in them.

Dashboards where every KPI has variance, targets, sparkline, segmented performance and missed opportunity numbers attached to them. For all 16 KPIs!

Consider this example that you'll find in your Facebook Analytics section...