Healthy Influence – Persuasion Blog

While I ran the Health Communication Research Branch in the Fed (okay, HHS/CDC/NIOSH/HELD/HCRB) I hit upon an idea I called a “persuasion engine.”  The web permits a different form of communication between source and receiver such that the source can monitor the real time behavior of receivers while they get the message.  Normally only face-to-face (f2f) interactions can do this and even that is difficult because of the cognitive load any interpersonal situation demands.  It’s hard to control yourself and accurately monitor the Other Guy.  With the web, you do not get as much information as from a live, interpersonal setting, but it is much easier to accurately monitor the Other Guy’s behavior.

My feral inspiration in 2000 with a Persuasion Engine was to capture receivers behavior as they process the message on a web page.  As you analyze this behavior, again in real time, you make adjustments to following web messages you show a particular receiver.  If properly done, this literally becomes an instance of self-persuasion as the engine provides more and more compelling persuasive messages with each click.  (You can also see this as an application of successive approximation from both For Me? reinforcement theory and an algorithm used to compute square roots.)

When I left the government two years later, I renewed my thinking on persuasion engines in websites and developed a couple of grant applications and hired a senior programmer to develop a new proof of concept program.  We never got funded.  Our scores were favorable, just not high enough to jump the cut line.  Then my writing work got more interesting and I stopped doing large project development.  The choices of life, you know.

I often read of other folks working on this problem, but not using the term, “persuasion engine.”  (Consider it mine.  Cite it as such or my team of lawyers will hit you like a pack of well dressed wolves.)  Today, for example, there’s a nice story about a research team targeting “thinking styles” of consumers as they visit web pages.  They’re in the ball park even without my really cool terminology.  I’ve seen others like this all aiming at modifying the message based on something the web user is doing.

However, I don’t think anyone is approaching the problem in quite the same way as a persuasion engine.

You need to do four things:

1.  Monitor the behavior of receivers.
2.  Classify it into persuasion functions.
3.  Provide new information consistent with your classification.
4.  Repeat.

What do you observe in the receiver?  Two elements.

First, monitor the mousing behavior of the user.  Everything that people do, everything, is a function of person and situation attributes.  How people mouse around a web page is a function of their personality and the situation.  Watch that and relate it to persuasion functions.

Second, provide choices on the same page that look different but lead to the same end point, then profile based on choice type.  For example, put a link to a new page on two locations, one as text, the other as an icon.  See whether the user prefers text or image.

How do you classify these observations?  How do you use the classification to design following web pages?

Everyone has their theory.  Try yours.  Relate mousing and choice preference to your theory variables in a large pretest sample of people like your target audience.  I’ve got my own theory preferences, but who knows?  There Are No Laws of Persuasion, right?

What applications are there for a persuasion engine?

Sell more stuff at EngulfandDevour.com.

Educate and train people.

Political advertising.

Just think about it.

This is likely to operate at the level of small effects (that 45/55 windowpane), so it is a long term, cumulative tactic, kinda like scraping kelp off of a sailboat during a long race.