Commentary

Deconstructing Words To Find Consumer Intent

by Laurie Sullivan , Staff Writer, August 12, 2009

What if you knew what potential and existing customers thought about your products and services? Could you pinpoint the exact day they might dig deep into their pockets and purchase that trip to Bora Bora or Madrid?

Consumer intent is extremely difficult to prove, even with the correct data to analyze it. Last week I wrote about BT companies merging in-store cash register data with online cookie data to better understand consumer buying trends. Let's take the concept one step further and add data that suggests consumer intent. The reality is closer than you might think.

OpenAmplify Tuesday released the tool, TopicIntentions, that aims to help advertisers and ad networks identify consumer intent through its computational linguistics engine, OpenAmplify v 1.1.

While BT platforms focus on where users go across the Web, OpenAmplify identifies what people read, positive and negative meaning of the content, and how the person reading the content reacts. It's done by monitoring blog posts across social networks and deconstructing the sentences that people write.

Think about Jacques Derrida's theory of deconstruction, which identified the ability to find the meaning of words by breaking them down to their core. Identifying consumer intent requires deconstructing the sentences by identifying subject, verb and object. Most platforms rely on general statistical techniques, such as how many times the word appears in the article or post, or what words appear closest to the subject of the content.

Understanding the relationships between words in a sentence allows the BT platform to target consumer intent. OpenAmplify CEO Mark Redgrave says the company is working with ad network Lotame to not only track people across the Web to monitor the sites they visit, but identify blog posts and the meaning behind them. The technology can monitor posts as people write them, which assists BT platforms to determine ads that might serve up alongside the written content.

Redgrave says Lotame is integrating OpenAmplify's technology. The two companies are evaluating the areas to tackle first.

OpenAmplify also has an online developer community where someone created a plug-in to the Drupal content management system, allowing people to amplify the content on their Web site. It lets them tag the content with semantic metadata.

behavioral targeting

5 comments about "Deconstructing Words To Find Consumer Intent".

Check to receive email when comments are posted.

Paula Thornton from iknovate, August 12, 2009 at 1:23 p.m.
Actually two companies have been at it for quite some time (admittedly, not cross-channel): Usability Sciences (WebIQ) and iPerceptions.
In the end there's still some serious wetware needed to bring it all together and synthesize it for 'consumable' artifacts (extracting the relevant 'story' and telling it). All part of the larger collection of Design Research to bring together the "context of 'intent'" http://www.fastforwardblog.com/2009/07/31/the-context-of-intent/
Reply

Jon Stamell from Stamell & Associates, August 12, 2009 at 2:02 p.m.

Cookie data, computational linguistics, blogpost monitoring, deconstructing sentences - it all sounds like a quick fix to try and determine what people really want. Why not ask them? Researchers have avoided this because they believe it's impossible to quantify coversational responses to open-ended questions. Not any more! It's not easy but perceptions, ideas, motivations and frustrations all can be quantified by Communitas Online (www.communitasonline.com) in massive numbers. Through CALCAT™ (Computer Aided Language Context Analyisis Tool), ideas are tagged, counted and tabulated to show what consumers really think. What's more, all you have to do is click on a data bar and see the actual verbatims appear. It brings research to life in the actual words of respondents so that our questions as marketers can actually be answered rather than trying to parse intent through counting cookies or bog posts. The best part about it is that it's common sense. Anyone wanting a demo can contact me at jstamell@futureshiftnow.com or jds@communitasonline.com

John Jainschigg from World2Worlds, Inc., August 12, 2009 at 3:05 p.m.

Finally, my happiness is complete - I've lived to see Derrida referenced in a blog post on behavioral marketing! Now I can tell my Mom: "You see? All that post-structuralist English-major stuff is now _job relevant!_" (grin)

I tend to align with Jon Stamell on this. Parsing commercially-useful intent out of people's (substantially presumed) history of interactions with deep-in-the-weeds internet content is not likely to move the dial in any substantial way, except in speeding the day when Firefox introduces session cloaking as a default.

Asking people what they want makes much more sense. Even better, give people an absolutely-flat, uninflectioned, un-gameable tool (not site, not experience, not brand envelope) for finding and buying stuff online.

Warren Lee from WHL Consulting, August 12, 2009 at 3:55 p.m.

Laurie, maybe I missed it, but do they do this without using any PII data? There goes the Privacy red flag. My first impression is that this is "one step too far" as far as BT goes, and agree with John and Jon that this will not move the needle in any substantial way. All in all, this is an exciting arena and I can't wait to see what else comes down the pipe. I am a big fan of targeted advertising's use to pay for content. In a perfect world, the tech companies would come up with a way to turn advertising into meaningful content.

Paul Knegten from Dapper, Inc., August 18, 2009 at 5:29 p.m.

Intent is great. But unless you have what to deliver to that intent (unlimited creatives, or unlimited offers), then what?