Judd SlivkaJudd Slivka
  • Blog
  • Work
    • Writing
    • Data Analysis
    • Teaching
  • About
  • Contact
Jun 7, 2013

The power of a map with words

Judd Slivka Data infographics, maps

Andrew Sullivan’s The Dish published a map originally published in the most recent issue of Psychology Today. They dubbed it “America’s Saddest Map.” It showed the results of a scholarly serious-minded and smart cartoonist’s analysis of the Craigslist “missed connections” sections.

Among other things the original analysis highlighted were the states where the most men seeking women ads were and vice versa. The map had each state’s highest-occurring keyword in it. The Dish chose to focus on this:

“Now look at the South – more people spy love at Wal-Mart than anywhere else, from Florida all the way to New Mexico. And that thread runs all the way through deep red America. Only Oklahoma cites the state fair as a mixer. The rest see each other under the merciless lighting of the giant super-store. This is how we fall in love or lust, where we flirt and look back: when we’re shopping. The big cities – like NYC and DC – showcase the random human interaction on the subway or metro. The Northwest has it all going on on buses.”

Yes, it’s another example of the coastal “Let’s crap on the Walmart belt” attitude. But there’s more interesting stuff here. Let’s take a look at those Walmart numbers.

If you look at the median income of all 50 states (the most recent three-year averages), you’ll see the first Walmart/Craigslist state doesn’t show up until all the way down at 32 — Texas, with a median income of $49, 195. In fact, 12 of the bottom 24 states in median income had Walmart as their highest-occurring missed connections word.

What does that say? It says there’s a concentration in the South, where wages are lower, of course. But Ohio’s in there and so are Idaho and Montana. Are there depressed wages there? Certainly in Ohio, where good paying manufacturing jobs have been lost and in Idaho and Montana, as well, where there is no base for good paying manufacturing jobs.

But it says something else. If you look at the states, you’ll see that with a couple  of exceptions — Ohio, Texas, Florida, Missouri — those 12 states are places with one significant metropolitan area and widely dispersed populations. Even looking at exceptions, you’ll see deeply rural states. Missouri has two major metro areas, but has a fairly dispersed population outside those areas; Texas has 2 1/2 major metros (Austin being the half) and a widely dispersed population, and it’s very similar to Florida and Ohio.

Ever been to a rural Walmart on a Friday night or Saturday afternoon? It’s the gathering place, a central place for the community. It’s the gossip center and a point of social concentration. In areas with dispersed populations, the  Walmart is placed in a regional commerce center of a few thousand — the biggest area around. And it draws for miles.

So when we get past the snark, the missed connection word study gives us a pretty good view of how people gather. Just like the highest occurring words in New York and D.C. and Washington and Oregon (all of which are dominated by single metropolitan areas with comprehensive public transportation systems) are about modes of transit.

  •  
May 23, 2013

Usability lessons from spring breaking

admin Data analytics, Crazy Egg, design, usability, Web

The good folks at web-usage tracker Crazy Egg asked five conversion experts to critique a PPC ad and the Universal Orlando landing page it led to. It was a massacre. There were comments about bait-and-switch techniques, bad colors, questionable layout choices.

All of them were true. But what Brian Massey at the Custom Creation Equation wrote resonates far deeper than just this page when it comes to web design and frequently made mistakes:

Test showing, not saying. This is an “EXCLUSIVE!” offer, but it doesn’t say to whom it is exclusive. Is it exclusive to previous visitors? Is it exclusive to people who have computers?

Try, “For our previous guests ONLY: A chance to come back and get your 4th night free.” If you can fly on a broom, show me. Don’t tell me.

Let’s get a copywriter in here. Is JK Rowling available? For example, what is the best thing about “Breakfast at the Three Broomsticks™”? It’s “one per person.” That’s as persuasive as lawyers get.

Get Voldemort in to do the layout. He doesn’t beat around the bush. Drop the navigation, the side bar, the trip planning video, and at least one of the three logo treatments on the page.

Get your call to action right. Is it “Book Your Trip?” or “Search?”

How would Hermoine translate “Need Assistance?” She would read it as “Are you lame?” Try “Call one of our knowledgeable Guest Consultants.”

Overall, this page is like Professor Snape: You can’t prove he’s bad, but you just can’t trust him.”

Setting aside the Harry Potter theme (the landing page was for a Harry Potter special), there are some excellent points here:

“Test showing, not saying” is brilliant and is a natural extension of the old editor saying “Show, don’t tell.”  Just saying something is “exclusive” or “low-fat” isn’t as effective a strategy as saying “for Harry Potter fans only” or “has 1/3 the calories of regular ranch dressing.”

Massey’s point about the graphic treatment on the page — it’s buried in a Voldemort reference — is dead-on. The page is busy and ugly and … overwhelming. It reeks of design-by-committee and a visit from the Good Idea Fairy.

The bottom line on this page is that there is too much there in there. And  it makes it harder for the consumer to make a sales decision than it should be .

Full article is here.

  •  
May 14, 2013

Exploiting the gray areas in a narrative

admin Story crisis communications, pr

‘And so it is written.’

It’s a fairly common sentiment, especially when we talk about the Internet Age. If it’s written down now, it can be found.

But what about if it hasn’t been written yet? What about, as so often happens, we have an intermediate ending and a vague notion of the beginning?  There’s that moment where the middle of the story hasn’t yet been written, and that may be the best place to influence the story’s path.

Think of the first 24 hours of reporting of the Deepwater Horizon disaster on April 20, 2010. We have a vague beginning (explosion) and an intermediate end (11 missing workers) which gives us a well-defined narrative arc.

But the middle part of the story was unwritten. There was nothing substantive about inspections, nothing about potential BP failings. There was not even any official acknowledgement  that there was an oil spill until April 24 and the first aerial photos of a significant slick aren’t taken until April 25.

Without lasting damages like an oil spill, this is a time where a company can use a gray area to bolster its own image.

What could BP have done?

— Told anyone who would listen that they would make good on any damages.
— Told anyone who would listen that the families of those 11 lost in the explosion and the 15 who were injured would never have to work again and that college for all their children was paid for.
— Loudly proclaimed that they were putting the best oil-rig firefighters in the world on retainer.

Some of this flies in the face of a traditional crisis communication approach in which you never take responsibility for an act until it’s either proven or painfully obvious that you’re responsible. But this is precisely the moment that you can take sincere actions that take advantage of the lack of information in the middle of the arc. That’s what lets you control the newscycle just a little bit longer and begin developing mitigating factors with the public.

  •  
May 7, 2013

Taxonomy: Big words, big results

Judd Slivka Data information use, taxonomy

You have a pot of information in front of you. A dataset, maybe. Or, more likely, a lot of data, not really in sets. You have to make sense out of it in order to make something out of it.

This isn’t just about data for databases (in fact, if you have a database, you already have basic groupings, by association). It can be in crisis management when you have multiple inputs of data coming in and have to organize them to be able to see the field. It might be quantitative. It might be qualitative. No matter what, you have to figure out how to organize it.

Which brings us to the most awesome thing about data: taxonomies. Strictly speaking, it’s how data is associated (red shirts vs. blue shirts, or shirts vs. pants; in a crisis management context, it might be “what I know as fact vs. what’s in Internet rumors,” or “medical reports vs. police reports”). When you give a group of datums — thus, data — a taxonomy, you give it a common home.

One of the problems of creating taxonomies, though, is creating rules. A database is only as good as its data structure and its contents. Lots of attention is paid to data quality within the cells of a database, but less is paid to the total organization of the dataset. If the structure isn’t well-defined and the rules aren’t thought out, then the data gets messy and it’s hard to get a consistent input.

Here’s an example of rules done right. It comes from Fringe Focus, a Chicago graphic artist who managed to sell 3,400 posters of all the ACME products from the old Wile E. Coyote cartoons in a Kickstarter campaign. In a post to his backers, he discussed his rules:

Second: Product choices

  • Any object that officially said ACME on it. Obviously.

  • Any product whose box, wrapping, or label said ACME on it.

  • Any product that appeared on an invoice, shipping manifest, slip of paper etc. that said ACME on it.

  • Labeled products that clearly arrived with other ACME orders. It can be assumed these were from ACME as well, even if their box did not read ACME in the name.

  • Any product with a named title on its box or label. Coyote ordered 100% of his items from ACME, so if it had an official name or box I just included it as an ACME item.

  • All books. The books accompanied or preceded ACME purchases, thus are assumed to be published by ACME as well.

Fringe Focus — real name, Rob Loutoka — does a fabulous job here of explaining what the parameters were. With clearly defined rules, input is simplified, streamlined and stands a much better chance of being clean.

  •  
What 84,521 clicks look like
Apr 24, 2013

What 84,521 clicks look like

Judd Slivka Data user experience, web analytics

What 84, 521 clicks look like, point by point.

Things of note :

1. The lower you get on the page, the fewer clicks you get, until you get to the footer, which challenges the traditional assumption of scroll bailout.

The middle of the page has feature content, rather than what we might call “news content,” or more directed entry points such as a map or a click off of a strong, specific image.

2. A strong footer saves the day. Web designers say this often, and when we built this page template I kept saying ‘If it doesn’t fit on the page, put it in the footer.’ Smart footer design allows the design to concentrate on what’s important and lessen clutter and it also rewards the down-scroll you get from someone who came onto the page serendipitously.

  •  
Apr 1, 2013

Mad monkey and why you need to monitor key terms, not just hashtags

Judd Slivka Data information flow, social

It’s getting so that you can’t bring a monkey anywhere without getting him quarantined.

The latest chapter in Justin Bieber is that his monkey was quarantined in Germany. But it lets us take a good look at the numbers behind how a story spreads and why you need to monitor terms and not just hashtags and handles in a social media story.

Because in the case of Bieber and his monkey, the number of mainstream outlets that picked up the story was far greater than the number of searchable/findable Twitter terms.

We put two keyword values into our search engine around 11 a.m. U.S. Central time this morning: 1) “Justin Bieber” and monkey; 2) “@justinbieber” and monkey . By 4:30 p.m., the first set had 280 results, the second set had 78.

Here’s the breakdown of where the keyword couplet of “Justin Bieber/Monkey” appeared:

One-third of the total pick ups on the couplet came from TV. Second was  Twitter, with 17 percent. Of interest here, more than 90 percent of that group did not have an overlap with Bieber’s Twitter handle. Far more people tweeted about him using his name than addressed him using the handle. An interesting note: Those who tweeted using his @ handle were relatively low wattage Tweeters, averaging less than 500 followers each.

The total direct circulation/audience of the outlets using “Justin Bieber” and “monkey” during a 5 1/2 hour sample period was about 22 million. The total reach of those tweeting the handle was no more than 6,000.

The lesson to be learned here is that you can’t depend on hashtag and handle searches. If this were a situation that were more sinister than a kid and his quarantined monkey, the real damage would have been done not on Twitter, but in the mainstream, big-bullhorn media.

  •  
«< 7 8 9 10 11

Recent Posts

  • Apps and gadgets, fall 2017 October 21, 2017
  • The mobile kit — Spring 2017 March 30, 2017
  • The apps I’m using — Spring 2017 March 29, 2017
  • The things we carry and don’t March 27, 2017
  • What apps do you use? The October 2016 edition October 29, 2016
  • Choosing the best rig for you — October 2016 October 28, 2016
  • My semi-developed Unified Theory of Translational Apps May 17, 2016
  • What’s the best mojo rig for you? March 29, 2016

Blog Categories

  • Data (14)
  • Idea Lab (19)
  • Media Matters (5)
  • Misc. (4)
  • Story (12)
  • Tools and Toys (29)
  • Uncategorized (1)

Portfolio Categories

  • Writing
  • Data Analysis
  • Teaching
  • Tags

    AirStash+ analytics apps BeastGrip camera+ class data design Diptic Doce Fire facts filmic pro Hyperlapse information flow interactive graphics iOgrapher iPad iPad Mini iPhone Joby LapseIt makayama mobile mobile journalism mojo Obama picplaypost Pinnacle Studio RecoLive MultiCam RØde Sennheiser snapseed social media Steller story storybox theory ThingLink Topsy Twitter usability Web word cloud writing Zimmerman

    ↑

    • Blog
    • Work
    • About
    • Contact
    © Judd Slivka 2014