Ah, Keyword Analytics. Analysts that have been using social listening tools with boo lean keyword logic for some time can tell you: this is easily one of the banes of our existence. The devil is in the details: swap out one “OR” for an “AND” accidentally, and you have a drastically different set of results, often with many of them being irrelevant. And, when you’re focused on your monthly mentions usage, having a rogue keyword can be more than annoying – it can be downright dangerous.
When I worked on the Agency side putting these queries together, I would often recommend to my teams to come up with a spreadsheet, so you can keep all the clauses of this complex Boolean argument straight. Despite our best efforts we would almost always see something get turned around, and have to correct our data sets as a result. So I presented this problem to our Development team, and as they’ve done so time and time again, they blew me away with an elegant solution!
So I’m happy to say that last month, we rolled out to our entire user base what they came up with: a visualization we call the Keyword Syntax Tree. You’ll find it when you’re entering Advanced keywords into your campaign, showing up right underneath the usual Keyword Validator. How does it work? As you type, we’re constantly checking what you’ve written to make sure it’s properly formed Boolean syntax, and as you complete valid clauses, the Keyword Tree renders what you’ve created in an unique visualization. For example, here’s a simple “OR” statement:
You’ll notice how we’re using both color and labels to visually describe to you the relationship between these two words, and how the colors in the Tree match back with those in the Validator. Now, watch what happens when I add an “AND” operator:
This time, the Tree has switched both colors and labels to show me the “AND”, and how it interacts with the “OR.” You’ll notice something else immediately – since I haven’t added parenthesis, the platform is honoring these operators in standard Boolean order. So in this case, it matches phrases that either contain “joe” and “jim”, or those containing “bob.” If I add parenthesis, you’ll see a different visualization:
See the difference? This time, it’s matching phrases that include “joe” AND “bob” or “jim.” Small addition, but totally different results. This Tree visualization works for “NOT” operators as well:
You’ll note how it’s honoring the Boolean operator “OR” I have within the “NOT” clause, and labeling that node with its correct color, but all of these branches are red because they’re all within a “NOT.”
We’ve already heard from customers that this is helping them to create smarter queries, and eliminating hours of tracking down where one phrase went wrong. In fact, I found it extremely helpful myself when building queries for our upcoming SXSW Parties study:
Having a visualization like this just makes complex queries incredibly easy, and gives you a powerful visual check to make sure the “shape” of what you’re asking for is correct.
We hope you’ll enjoy the Keyword Syntax Tree yourself! As always, feel free to drop me a line if you have any questions!