7 Data Presentation Tips: Think, Focus, Simplify, Calibrate, Visualize++

elegantThere are three elements to our "big data" efforts, or unhyped normal data efforts: Data Collection, Data Reporting, and Data Analysis.

(More on that here: DC-DR-DA: A Simple Framework For Smarter Decisions .)

We are all aware that the best companies in the world have an optimal DC-DR-DA allocation when it comes to time/money/people: 15%-20%-65%.

All well and good.

But there is one crucial part we often don't invest in sufficiently. The last mile. Data presentation! The actual output that is almost singularly responsible for driving the change we want in our organizations. The thing that is the difference between an organization that data pukes and the one that influences actions based on understandable insights.

I believe we should present our data as effectively as possible in order to first build our credibility, second to set ourselves apart from everyone else who can present complicated graphs/charts/tables, and third allow our leadership teams to understand the singular point we are trying to make so that the discussion moves off data very quickly and on to what to with the insights.

A vast majority of occasions where data is presented (reports, executive dashboards, conference presentations, or just plain here's a automated emailed thingy from Google Analytics ) end up being abject failures because most of the discussion is still about the data. And if you are sitting in a Nth level tactical meeting, that is ok. But if the occasion is a strategic discussion, any occasion about taking action on data, then you need to get off data as fast as you can.

It is hard to do. After all you spent so much time on collection, reporting and analysis. You want to show them all data stuff and how much you worked and how cool your technique was. But trust me, it is better for your career (and, this is a lot less important, but much better for your company/audience :)) to get really, really good at data presentation.

This post shares eight before and after examples that illustrate seven data presentation tips that I hope will inspire you to look at your report/dashboard/PowerPoint slide in a new light. We will look at some simple errors, and some much more subtle ones that end up limiting our ability to communicate effectively with data.

Here's a quick summary:

We are going to have a lot of fun, and learn some not-so-obvious lessons.

It's not the ink, it's the think.

An important point first.

This post is not about tufte'ing your work. It is not a post about expressing your inner Excel geek with the most advanced remastered sparklines or conditional scatter plots. Advanced, sophisticated visualizations are important. But I find that so many times people focus on the ink and not the think. Hence all the insights-free data visualizations floating around the web that are totally value-deficient, even as they are pretty.

In this post I simply want you to focus on the think and not the ink. What was the error in thinking? How can you ensure you never make that error? Then, go express your inner visualization beast. :)

[My inspiration for a focus on the think: Bob Mankoff]

Lesson 1: Don't be sloppy. Your data presentation is your brand.

This graph is from an article by the consulting company McKinsey.

It actually shows very interesting data. The article is a bit dry, but valuable.

Yet, I could not get over how sloppy the graph was. For me, and perhaps for others, the sloppiness made the data appear to be an amateurish effort (surprising, given the source) and took away from the deservedly mighty McKinsey brand.

Can you see what the problems are?

email over social media 1

The first problem is that the title is weirdly placed. Then the y-axis legend is even more weirdly placed. The most important part seems to be to get the names of the company, gigantic, over two lines and distracting.

Finally, this is picky, but why is most of the x-axis yearly and then suddenly just until Q2, 2013? And if it is only two quarters of data, why is it taking up the same distance as represented by one year?

Surprisingly sloppy from McKinsey, right?

Watch out for these errors. People in the room (in a small room or a board room or a conference auditorium) will know a lot less about the data than you will, their first impression, and often the lasting impression, might be how clean your data presentation is.

Even without access to the raw data (let's say I'm a busy McKinsey blog post writer), you can make a couple of simple changes to the graph to make it cleaner and less sloppy…

email over social media fixed 1

Clean up the title, rephrase it.

Move the y-axis description to the right place.

Make the source attribution much smaller. If the data is good, people will seek it out. If the data is stinky, no one cares. Either way, why make it intrusive?

Scroll back up. Then down. Much cleaner, right? 30 seconds of work.

If I had the raw data, I would also fix the x-axis and representation of the partial 2013 data. That is still bothering me. But at least you can see what 30 seconds can do.

When it comes to your work, take the 30 seconds.

[PS: The data in the graph is cool, you can see my brief analysis on my LinkedIn Influencer Channel: Email Still Rocks! Social, Surprisingly, Stinks!]

Lesson 2: Bring insane focus, and simplify.

I'm sure you've either seen someone present a slide that looks like, or you've created a slide/executive dashboard like this one. Or, both.

: )

Before you scroll any further, what errors, subtle or obvious, do you see? Don't rush. Give it some thought.

cpc trending brand non brand 2

[Minor Rant: Never, ever, never obsess this much about CPCs. Yes, cost per click is metric. But if you had to obsess about something, obsess about the value delivered to the business. You will never obsess about the cost per trade of your E-Trade portfolio, right? It could go down from $10 per trade to $1, and you could have completely gone bankrupt as a result of your trades. So, don't obsess about CPC. Focus on Economic Value from your search advertising. Focus on Profit from your search advertising. Focus on the outcome. As long as you make a profit, does it matter if your CPC is $1 or $200? And would it matter if your CPC went from $200 to $1 if you were making no profit?]

The metric CPC aside, we do present data like this all the time.

The first challenge is that there is too much of it. We have actuals and we have the YOY change. Then we have it for the company and its category. Finally, we have it segmented into desktop and mobile and as if that was not joyous enough, further segmented into Brand and Non-Brand.

As if that was not enough, the data presentation itself is a bit uninspired.

We can quickly fix it though.

First pick one primary thing to focus on. When you design dashboards this is absolutely critical.

In this case, I believe, the most interesting thing is the YOY change. I bring it center stage, and make the actual CPC as small as I possibly can (in case someone wants it that desperately).

Next I create a simpler data presentation, God bless Excel, by creating two big clusters next to each other. Now it's just a matter of two similar columns that we can distinguish with the use of color.

Here's the result…

cpc trending brand non brand fixed

Again, something very quick you can do. (I'm sure like me you have a favorite custom font you use to make your presentations really yours.)

The orange and purple are easy on the eyes, and distinguish the two clusters nicely. The size of the font used makes the things that should stand out, stand out easily.

Notice because the company performance is all in one row, it is much easier to see that their CPC year-over-year change is less than the category (something harder to see in the original version).

Bring insane focus to your data presentation. If you can, focus on a singular metric for each module/slide/element. Then present the data as simply as you possibly can. And often, you don't need to go very far from the defaults in Excel – though you are welcome to use any software you want.

Lesson 3: Calibrate data altitude optimally.

Here's a more subtle error.

Ignore the ugly graph and the terribly formatted axis, time periods used, etc. All simple fixes.

Look at the text under the graph. Do you see the problem? Don't scroll any further. Look at it again, see the mistake made?

confused paid organic 1

It is not completely obvious, but the Analyst is expecting that in the very short time the leadership team has to look at this data, that they'll also be clever enough to do the math for each row, commit it to memory and then compare all four rows and figure out which video is performing better.

Terrible error in judgment. The altitude is all over the place!

You are the Analyst. You do the math. Then make the hard decisions and figure out how to present data as effectively as you possibly can.

In this case I had to decide what the key point was (this is the think part). I believe it was that using advertising to drive views of a video fueled organic views as well.

That gave me the anchor, paid views. Then it was simply a matter of figuring out the best way to present the data. I decided to use an index of 100. All that's left now is to do the math in Excel and paste it on to the dashboard…

confused paid organic fixed 2

The recipient can get to the insight really fast because there is less data (fewer words and clutter), it is well thought out, and we can move to asking hard questions about performance.

What the heck happened with Video B? And OMG what is up with Video D???

That is what you want, shift the discussion from the data to what happened and what to do now.

Bonus: As the smart Analyst that you are, at this point you'll realized Earned and Paid Views don't tell the full story. So you'll change the table to Total Views and % Earned. You would not have known that's what you needed if you'd stuck with your original textual version! The value of focus and think.

Lesson 4: Eliminate distractions, make data the hero!

Raise your hand if you've not created a slide like the one below for your presentation. Come on!

My hand is raised.

We have all done this.

And it is so silly.

We take the most interesting part, the data, and surround it with clutter that only makes it harder to understand what the point is. The data is the hero, what is the need to have the arrows and the box and the descriptions? Is there any need for the useless stock photos (and what is up with the magnifying glass to represent research, who does that?)? And why repeat "use online sources," is that not obvious in the awfully crafted title?

Look at the image for a moment. Don't scroll. Stop. Really. Don't scroll. How would you decrapify this slide?

Got an answer? Ok, now scroll.

research to purchase process 1

Share your decrapified version via comments below.

My process was to simplify the title to something more direct and easy to understand. Then use three different bars to represent each stage of the process, and to fill each up to represent the percentages. Finally, I'm slightly allergic to terms like awareness and consideration. They are too generic, they encompass too much. So I took the direct route, just wrote down what each bar actually represents.

research to purchase process fixed 1

You can use different colors, mix to suit your own taste. Red in my case is to make the online usage stand out on a very large screen.

I'd experimented with having a break in the gray x-axis (yes, I worry about those things!), it looked nicer. But visually it ended up representing a break, rather than the continuity that each stage represents. Hence the single line you see above.

If you spend sometime on the think , it is so much easier to decrapify the data presentation to focus on the most essential element and make data the hero (again, so that you can get off the data very quickly and have a discussion about what the business should do).

Lesson 5: Lines, bars, pies… stress… choose the best-fit.

If you are a student of the Market Motive web analytics master certification course, you'll note my love for segmented trends rather than snapshots in time when it comes to data presentation.

Trends are often better at delivering deeper insights. And because all data in aggregate is crap, segmented trends are even better!

But, as all smart analysts know, often is not always.

Here's a great example… The dashboard module shows how American's consume media, and how that behavior has changed over the last four years.

Please take a minute and reflect on the graph. Do you love it? Does it communicate the change optimally?

line graph us media consumption 1

You'll agree, the graph is nice and clean. It is easy to understand what is going on. Sure we can line up the numbers on the right correctly, but that is a minor point.

As a Digital Marketing Evangelist, you can imagine I love the data. : ) I was not sure that I love the line graph.

I felt it would take too long to understand just how much things had changed. People would spend too much time trying to understand the graph. And even then, at a deep gut level, not internalize it (even though to you perhaps it is utterly obvious).

My decision was to eliminate the trend. Except for TV, the trends adds almost no value (and even for TV just a little). This allowed me to switch the x-axis to each media channel, they were the heroes here. And finally, switch to a bar graph.

Here's the result….

bar graph us media consumption 1

I believe this version shows the change much more starkly and since you can look at one channel at a time, you can absorb the change much, much faster than with the line graph.

While with the line graph you could see people spent more time with digital than with TV in 2013. The big rise in digital consumption vs. 2010 is much more obvious now. And while TV is physically from Digital in the above picture, you can easily see that one is much higher than the other.

Remember, often is not always. Question how you've always done things. Even question your teacher who might love segmented trended graphs! : )

Understand who your audience is, think about the point you are trying to make with your analysis, and then use the best-fit data presentation method.

Lesson 6: Consolidate data, be as honest as you can be.

This example comes from a presentation. The data was spread over two slides. Notice how nicely it is presented.

The first slide showed the desktop and laptop performance for search traffic for puppies (real data below, just not that category!)…

searches for puppies desktop

It is easy to see how puppies are doing in context of the average number of searches for land animals and sea animals. Put another way, company performance compared to two benchmarks.

The second slide illustrated the mobile search performance for puppies, and compared it to the same categories…

searches for puppies mobile

Both sets of data presented simply. You cannot misunderstand it.

So, what is the problem. Look at the graph above carefully. Then scroll up a little more, look at the first one. Now scroll back down.

See the problem?

One obvious problem is, why spread the data on to two different slides? Most people are terrible at keeping track of things as they jump slides/pages.

The second problem is more subtle.

The graphs make it seem like there are two similar sized problems to deal with for us as PuppiesRUs Inc. But that is not really true. Look at the y-axis.

Perhaps, for a good reason, we want the company to believe that they are similar sized problems because our company sucks at mobile and we want to light a sense of urgency under our collective butts.

I believe as an Analyst we should be as honest as possible in these cases. (I'm NOT implying that there was a deliberate attempt to not be honest above.) We should show the data in as honest a way as possible, we should be as objective as possible.

I simply took the data in the two graphs and put it on to one graph, same bar graph, and fixed the title to make the presentation simpler (I hate long complicated titles).

In an attempt to pay an homage to the importance of mobile, changed the color to red…

searches for puppies fixed

To our leadership team, the recipient of our presentation, it is really clear how we are performing overall and in mobile.

It is also clear that desktop plus tablet, blue, is the most important area of focus. We have to keep the pedal to the metal when it comes to that. But that mobile is also an important area deserving some dedicated focus.

There is no chance that they will inadvertently think the size of both the opportunities is the same.

An effective presentation of data by 1. consolidating it and 2. having it play off the same y-axis.

Lesson 7: Ditch the text, visualize the story.

Often we hear that data is overwhelming or that graphs are evil or that tables suck or… well, I'm sure you've heard it all.

Our response to that is to try and "simplify the story" by eliminating all that and just writing the insights in text with a big summary number.

That strategy does work some times. More often than not you end up with something super-ugly and value-deficient like this…

search data puke 1

Imagine yourself to be sitting in the audience and trying to internalize everything that's going on here! I'm sure someone is going to walk you through it. But still. Do you think there is any chance you can grasp the multiple agendas at play above?

I seriously doubt it. Scroll back up. Look at it!

Even if you only have two minutes, all I had in this case, it is pretty easy to fix the above textual representation and make it much easier to understand what is going on.

First, get your custom font. Ok, kidding.

First, think of what the key point is and replace the long red-book ended title with it. In this case: Search Opportunity.

Then draw a bar in PowerPoint, eyeball the size (no, really, don't even go in Excel to create the graph, no one is going to notice!), and fill in the sub-components.

For data you can't find an obvious home for, use call-outs.

Two minutes later…

search simple data presentation 1

So much easier to see that story is about how many people search for our company topics and that weight management and monitors are the most interesting. In this case we have the data that can fill out rest of the bar, but we want the leadership team/audience to focus on just two and those are the ones you see above.

It is less obvious how to illustrate the mobile growth. Two more bars? Perhaps a heat-map showing high and low? Nah! Just add two call-outs and you are done!

When the data's end state is a PowerPoint/Keynote presentation, use the fade transition (all other transitions are evil) and bring one piece of data at a time up on the screen. It will look beautiful and the audience with stay with you as you narrate other insights you know that are not represented on the slide. [A style of presentation you should use every time you present anything.]

Here's another example of eliminating text, reducing complexity, focusing the the key point and visualizing data simply to get off the data quickly and discuss actions.

Pause. Look at the example below. What is done right or done erroneously? If you had to improve on the power of communication for this example, what would you do?

Pause. Really think about it. Got it? Now scroll.

media targeting efficiency

The first simple mistake you likely won't make as an analyst is to use two different things to represent the same number. For example, either stick to the dollars or use the percentage. This might not seem like a big deal in isolation, but every little bit like this takes a tiny bit of your credibility away and it causes the audience to have to shift their minds a little. Over a number of these types of mistakes in your dashboard or your presentation take away 0.25% here and 0.5% there and 1% somewhere else. Taken together, you lose 30%. Why dig that hole for yourself to have to climb out of?

The second simple mistake, obvious in hindsight I'm sure, is that there is simply too much text. Why not simplify the data presentation to make it boom (!) impactful right away?

I did like the map, but it was intrusive. So my first act was to take the map, fade it out (use a white transparency, 13%). It is there, but it is not in the way.

Then I did not like the numbers, they don't add any value. Just throw in two simple bars (standard shape in PowerPoint, no Excel necessary), and add a touch of color to show targeting efficiency of TV and Radio. Finally add the bridging text and use the brace (use the little yellow handle to drag the brace so it is aligned) to show how well or badly each media channel is doing.

Red is bad, blue is good….

media targeting efficiency fixed

Scroll back up. Then back down. Then up. Then down. (Think of the Old Spice ad! :)

The presentation is simpler. Even without reading anything you can get a sense for what is good and bad. The questions will come fast and loose: Why do we do TV? And if there is 75% leakage, is it still worth it? What is the optimal media-mix for our efforts?

We believe that summarizing our findings in text is the solution. We believe tables and graphs add complexity. We could not be further from the truth.

Closing Thoughts.

It's not the ink, it's the think.

It takes a tiny amount of time to really look at the data you are presenting, really think about what you are trying to say and identify the singular point. Once you know that, it is only a couple of minutes of work to decrapify the report/dashboard/slide/spreadsheet and ensure we are presenting data as simply as possible using the most optimal visual.

You worked so hard to collect the data. Then invested all that time and energy in reporting it. Finally, really dug deep, did the analysis. Don't stop there. Spend time optimizing the end product. Your goal: Get of the data as fast as you can, switch to the discussion of actions.

Victory, I promise, will be yours!

As always, it is your turn now.

Which one of the eight examples above is your favorite? And the least? Would you have taken a radically different approach on any one of them? Care to share your version? What are your go to filters for taking something complicated and making it simple? What is your favorite annoying data presentation method? Is there a visualization strategy that consistently helps you switch the discussion from talking about the data to talking about what to do with the insights?

Please share your insights, recommendations, critique, alternatives and complaints via comments.

Thank you.

Comments

  1. 1

    Hi Avinash,

    I was looking for some ideas on visualizing the data with the clear story. Fortunately your blog post bang on during the same time. I am happy to read the post and will definitely apply these techniques in my work.

    Thanks,

    Bhagawat.

  2. 2
    AJ Bellarosa says:

    Hello Avinash,

    Yet again the content you deliver to the digital analytics community is insightful and adds so much value to the work I deliver to my clients on a daily basis.

    As I am right in the middle of adjusting weekly and monthly reports for a global brand's Radian6 instance, your perspective re: data visualization makes a lot of sense.

    I'm wondering what your thoughts are re: including data points in graphs? Too much clutter? Or a necessary inclusion?

    Thank you!

    AJ

    • 3

      AJ: It is hard to comment without seeing your graph, but usually I find that data points clutter the story that we are trying to tell.

      If the graph is done well, you might not need the data points.

      If you want to share your graph with me, I can take a look at it and share a specific recommendation.

      -Avinash.

  3. 4
    Tim Wilson says:

    Nice post, Avinash!

    I'm amazed (dismayed, sometimes, depending on my mood) how much energy I spend trying to coach analysts to put more effort into the presentation of their results. I have bought or encouraged others to buy Stephen Few's "Information Dashboard Design" more times than I can count (he has a new edition out). Few provides a number of directly applicable principles, backed up by a light dose of neuroscience to explain both *why* this stuff matters as well as very practical tips/concepts for making analysts better presenters of information.

    That includes "minimizing the data-pixel ratio" (which he lifted from Tufte's data-ink ratio), the fact that we can only hold 5-7 discrete elements in our working memory at one time, how much easier it is for us to process size/shape than numbers, what percent of the population has some form of color blindness (9% of U.S. males…and why that matters when presenting visualizations), and more.

    A number of your examples here touch on some of these concepts. And, some of them could be further refined/improved (which is par for the course — I once asked Ian Lurie after a presentation he did on this topic if he looked back on work that he'd presented a year ago and immediately saw things that he could have done to make the visualizations of the information more effective, and he immediately responded, "Every time!" I have the same experience… and every analyst who is appropriately concerned with that final step of an analysis *should* have that reaction!).

    • 5

      Tim: Wonderful advice, thank you for sharing it with us.

      I can also empathize with Ian's reaction. In preparing for this post I looked at my work over the last few months and I think nine times out of ten I found at least one small change I would make, usually a big one!

      We live, we practice, we learn. : )

      Avinash.

  4. 6
    Shailesh says:

    With the easy availability of graphing/charting tools (like Excel) we usually fall into the trap of working with ‘defaults’.

    However, like you said, give an extra few minutes and the presentation/comprehension value of the same chart/graph can be greatly enhanced. The trick is to obsess less about data perfection and devote a bit more time into data presentation. That’s not to say that you compromise on data quality or accuracy, but simply to put a little more thought into how you present the results.

    It’s amazing how good visuals convert even the most die-hard “I want data in tabular format” type of stakeholders.

  5. 7
    Rebecca says:

    Loved this post, as always. I guess my favorite tip is the third one.

    Managers usually tend to believe they have this need to see "all the data" and "all the numbers" and know "everything". But in fact they don't. And when you're an analyst, most of the time you don't get to say what's best for them.

    One of my most repeated mistakes over time was allowing this obsession to go on. I started realizing I was showing the same numbers in different dimensions with the same relevance. For instance, I used to show a trend, create and axis with ordinals and then add percentage. But this is redundant. And the worst thing is people who saw it liked it. Moreover, I noticed I was taking a few minutes to explain the slides, instead of just passing by them, something that shouldn't be happening at all.

    So I started to take some time (it is not about just 30 seconds for me) and think about creating a story with my slides. I admit it still is the most difficult thing for me: doing the math myself, presenting only what needs to be seen and letting the story speak for itself.

    Thanks for sharing the tips. =)

    • 8

      Rebecca: That is a brilliant insight! We get caught in the trap of giving our senior leadership what they want.

      I've found that if your audience is asking you for more and more data (and to make your dashboards and presentation complicated and "data rich") there are two things that are happening:

      1. Your leadership does not trust you and your skills. Perhaps because you are new or they don't have confidence in you. Then they want to know more and more because they want to be extra sure. Try to earn their confidence in this case.

      2. The insights I'm delivering based on the insights are not good enough. Because I can't "distract" them with the good insights, valuable actions to take, they are left with just looking at the data and critiquing it. Find better insights in this case, spend more time on that.

      Thank you for sharing your fantastic comment and letting me share these two tips.

      Avinash.

      • 9
        Rebecca says:

        Exactly! That's the whole point! After I started taking some time to think "what am I gonna say that'll be absolutely killer?", "what do they wanna know when I show these numbers?", "where do they want to get when finding about this"?, it stopped happening.

        Now I can do my analysis , get to the point faster, work around theories with the slides and, as you said, tell a story through math. =)

        It was difficult though. At first, all you think about is "omg, critiques again". I had an awesome "Assertiveness & Communication" course and it sure helped a lot.

        For those that still have to deal with it, I mostly recommend it. =)

  6. 10

    'Oh, people can come up with statistics to prove anything. 14% of people know that.'

    – Homer Simpson

  7. 11
    Jess says:

    Great post!

    However I think font selection is important. I know you want the charts to look better, but your chosen font is not a good choice.

    Poor readability at small sizes, uneven kerning, and a general noisiness makes it a less professional choice.

  8. 12
    arjun says:

    Hi Avinash,

    The data in the first graph showing the acquisition from which channels the conversions happen more can help the advertiser to concentrate on the channels which brings more conversions.

    The data and visualization is help a lot.

    Thanks for sharing the data.

  9. 13
    Matt says:

    Great post, and love the recommendations at the end to take a step back and evaluate the work to make sure the core points from your time intensive research efforts really shine through and move the conversation forward from research to action.

    Not to pick nits, but the missing legend in Lesson 6 can be just as much a distraction as the McKinsey x-axis issue from Lesson 1 – and at least always leaves me saying 'what am I looking at?!'.

    • 14

      Matthew: You are absolutely right, the legend is very important and should always be included.

      I end up creating most of my visuals to present as a part of a keynote I give. In those cases I even remove the legend because I'll use transition and you'll see the blue bars come up first and I'll talk about them, then I click and the red bars will come up and I'll talk about the contrasts. But because I'm introducing each element there is no need for a legend.

      But posted on this blog as a picture, the legend is sorely missed.

      -Avinash.

  10. 16

    This post could not have been published at a better time pour moi.

    First, I had too much fun laughing and reading this post on a dull Tuesday afternoon. Second, I have many daunting yet exciting analyses in the pipeline due for next.

    This post is making me wish that much more to fast forward data collection and accuracy and get to the fun part of presenting the data with actionable insights. Data presentation is one important skill I have identified and acquired as a Market Motive student in your course. Excel is ok. Powerpoint is awesome.

  11. 17
    Grace Lau says:

    I really, really liked today's blog post. Thank you!

    My favourite example – % of Ad Spend that reaches in-state audiences only. Love it!

    My least favourite example – The Digital's Role in the Research to Purchase Process

    * Why use a stack graph when you can use a regular one?

    * I kept looking at the blue, green, and yellow bars thinking, "What are they for?"

    Still, this is probably my favourite blog post!

    • 18

      Grace: I ended up using the stacked bars because I wanted to denote that they are three completely, if related, things. Using the same color would in a subtle way imply that they are the same people going down the "funnel."

      But they are not. In fact more people use the web for research, and not everyone in the 40%, purchase, might have been in the 52% and 44%.

      The blue, green and yellow bars, hopefully say that.

      There is, of course, no perfect way to do this. Just many ways to get closer to clarity. :)

      Avinash.

  12. 19
    Brenton Jones says:

    Great post Avinash. This topic is very close to my heart. I can talk first hand to the importance of how to best present data at the strategic executive level.

    Once in a previous role I paid heavily for not taking the time to sell my findings properly; raised more questions than decisions! It costs me close to 6 months delay in getting approval to invest in further website optimization work. Didn't happen to me again.

    It pays to think creatively in how best to sell the story. Most valuable lesson by far for me is: Lesson 7: Ditch the text, visualize the story.

    Thanks again!

  13. 20
    Pradip says:

    Hello Kaushik,

    I would like to add a mindset approach in addition to the techniques you've illustrated.

    One good mindset approach for someone presenting data is to assume himself / herself is the one who needs to make the decision. Then ask yourself, what / how would you like data to be to make the decision?

  14. 21
    Josh Braaten says:

    My favorite was the ad spend per medium. My gut told me the first couple years were irrelevant but not until you moved it to a bar graph did I realize what to do with that insight. It completely changed the story.

    On the puppies search graph, the only thing I didn't quite like is not knowing desktop vs. mobile. Are you opposed to legends and if so, what's your recommendation to give context on what red and blue mean?

    My favorite tip is that the "star" of the story should always be your x-axis. Great post!

  15. 22
    Jai Rawat says:

    Great post Avinash.

    I liked all the examples but disagree with your approach for #5. I think the fact that time spent on TV has declined from previous year is a very important bit of information that got lost when you converted the trend graph into bar graph.

    One of the best data visualization I have seen was in a TED Talk by Hans Rosling. It blew me away. ted.com/talks/hans_rosling_shows_the_best_stats_you_ve_ever_seen.html

  16. 23

    hmm… This post seems to be more "depends" on data post rather than clear cut instructions. I guess visualizing is more person dependent (Except when it is outright wrong!). For example, in lesson 5, I found the line charts to give richer information than the bar graphs :-). Is there any best practices doc that you know of on – how to choose the right chart type for your data?

  17. 26

    Great post as ever, Avinash.

    Your concluding remark is spot on: 'switch to the discussion of actions'. There is so much conversation and buzz around big data, but often the amount of data is used to cloud the fact that only a limited percent of the data is actually useful/actionable.

    I've been reading Nate Silver's 'The Signal & The Noise' book recently, and he often remarks that the best analysis often comes when you can focus in on a few key metrics/sources, blocking out the 'noise'. To bring this back to your main point, taking this same approach of stripping away superfluous information will lead the most useful presentation of data.

    I work mainly with PPC/AdWords data, and have trying lately to use the R language to better visualize the data (see my thoughts here: http://uk.queryclick.com/seo-news/visualising-adwords-data-using-r/). Whilst still rough in presentation, the tenet of 'It's not the ink, it's the think' has proved to be very beneficial and a bit of a revelation for me.

  18. 27

    Excellent analysis Avinash, I'd love to see how we can track multi device conversions.

    Say you start a search on your phone then you convert when you're back home from your computer. How can we track that and attribute both cost and conversion?

    Thanks,

  19. 29
    Aliza says:

    Wonderful post and great timing!

    I've been working on improving my own presentations and this has really come in at the right time. I tend to do a lot of trend graphs and your 5th example using bar comparisons instead of a trend line was my favorite. I also feel like a legend should always be on a graph.

  20. 30

    Great post with great examples and tips, as always.

    I always thought the data visualization folks go too far in arguing over pie charts versus line charts – usually pie charts are considered non grata – and forget to think about the core message to be delivered; the insight you got from the data that you wish to get across.

    So this was a great 'grounding' in simplifying things to just what is necessary.

    Thanks
    Paul

  21. 31
    David Sealey says:

    For those who want further insight on presenting data for reports I highly recommend Jon Moon's book IMPACT. It has some really good worked examples in and includes some great information on rounding in reports to clarify data.

    David

  22. 32
    Erik Feder says:

    Great post as usual.

    Just one question – on #6, you wrote "It is also clear that desktop plus tablet, blue, is the most important area of focus."

    In your version of the graph, it is not stipulated which color represents mobile and what the other color represents. It's obvious to me that the other color represents 'not mobile' i.e desktops/laptops, but in the name of understandability/simplicity isn't it important to have a key showing the viewer what our data represents?

    (PS – sorry if this is annoyingly nitpicky, it really is an awesome post)

    Erik

  23. 33
    Erik Feder says:

    Never mind my last comment – just saw that you already addressed this with Matthew.

  24. 34
    Eli Kallison says:

    Wow, what a great display transformation in #7 for targeting in-state consumers more effectively.

    Love the way you did that bar graph.

  25. 35
    Gautam Gogoi says:

    Hello Avinash,

    Super blog. Truly enjoyed reading it.

    I was making a deck showcasing a new idea and came to this site for inspiration. But i noticed that what you have wrote here is mainly with regards to data presentation. But since i was making a pitch for an idea this was not entirely relevant because i dont have data points at this moment. The point i am trying to make here is that as analysts we also have to make presentations without numbers regarding ideas we would like to implement. So such kind of presentations are also equally important.

    So if you agree you might be kind enough to take that up in one of your future blogs. Would love to hear your views on the same.

    Thanks

    • 36

      Gautam: Most of my public speaking is in context of presenting new ideas and frameworks that influence thinking in different ways. I completely understand your point.

      I'll definitely try and consider writing on the topic in the future.

      Avinash.

  26. 37
    Andy Melton says:

    I really enjoyed this post and your take away points are spot on.

    Throughout this post I kept remembering something I learned as a young lieutenant in the Navy a hundred years ago at the Pentagon when I had to present data to decision makers. We used to say, "Be very careful about giving an admiral a number. He'll just use it."

    I have always felt responsible to ensure that the data I presented would lead to a decision the analysis supported and that confusing presentations can lead to erroneous decisions.

    • 38

      Andy: Thanks so much for sharing your experience!

      I love the quote. Something we should all of us in the business of data should print and keep handy. :)

      Avinash.

  27. 39
    Oscar says:

    I like this part on your article: Eliminate distractions, make data the hero!

    Sometimes we tend to write a lot of things and lose our focus on interpreting or explaining more about the data presented.

  28. 40
    BP says:

    Well i am not so good in making presentations for my business, But what an article you have shared with us!

    I can easily learn and apply all the tips and tricks on my next presentation.

    Many Thanks.

  29. 41
    Luis says:

    Thanks for the nice post Avinash!

    I'd love to know a bit more about data presentation tools. Which ones do you use?

  30. 44
    Big Data Mama says:

    Great tips for visualizations!

    Do you work with interactive or animated charts? What are your thoughts about that? Do you think interactive charts/graphs add value for the end user in revealing more insights, instead of a flat hard copy visualization?

    Everyone wants "clickable" data with more layers or information beyond a static report. My favorite example you gave was consolidating the graphs into one easy-to-compare slide ("Puppies Search Traffic").

    I like your blog. I am going to check out your Twitter page.
    Thanks!

  31. 45
    Paul says:

    Thanks for the insightful post. Ive alsways been a fan a Prezi, if the data is layed out ok it can really make it look good and appear more coherent.

Trackbacks

  1. […] 7 Data Presentation Tips: Think, Focus, Simplify, Calibrate, Visualize, http://www.kaushik.net […]

  2. […]
    I really enjoyed today’s blog post from the famous Googler and entrepreneur Avinash Kaushik on tips for data presentation, particularly the importance of moving discussion quickly off data and onto insights and actions.
    […]

  3. […]
    Does your brain check out when charts and graphs pop up on a screen? The versatile Avinash Kaushik teaches us how to make a lasting positive impression, even when presenting otherwise boring data. There are plenty of helpful (and illustrated!) tips in 7 Data Presentation Tips: Think, Focus, Simplify, Calibrate, Visualize++. Data doesn’t have to be a cure for insomnia!
    […]

  4. […]
    7 Data Presentation Tips: Think, Focus, Simplify, Calibrate, Visualize
    […]

  5. […]
    I intend to write a longer post on dasbboards for direct marketing, but for now I’d like to share a great post on data presentation tips: 7 Data Presentation Tips: Think, Focus, Simplify, Calibrate, Visualize
    Avinash Kaushik is a data guru and a data driven marketing thought leader. You should follow his blog!
    […]

  6. […]
    Link here: http://www.kaushik.net/avinash/data-presentation-tips-focus-think-simplify-visualize/
    […]

  7. […]
    A Piktochart infographic made with the information from 7 Data Presentation Tips. 
    […]

Add your Perspective

*