Skip to main content
car gizmos

The autoLYZER was launched on Facebook on May 23 and is a Canadian-only pilot project that takes search to a different level by matching behavioural patterns and data from the social network and matching vehicles to that.autoLYZER

It's fair to say that, as consumers, we buy cars that would fit with both our budget and personality. AutoTrader has attempted to automate the search phase in that process by mining a user's Facebook data as a way to determine the rides deemed to be the best fits.

The autoLYZER was launched on Facebook on May 23 and is a Canadian-only pilot project that takes search to a different level by matching behavioural patterns and data from the social network and matching vehicles to that. That includes location, the pages a user "Likes", marital status, level of education, gender, number of friends, comments on pages and photos and memberships. These all come into play in the analysis.

To find the right vehicles, the autoLYZER trolls through 500 different makes and models from 50 auto makers available on the AutoTrader.ca site. Rather than just suggest three cars, the results turn up vehicles that are actually available in real-time on the site. To allay any privacy concerns, AutoTrader says it doesn't store the data it analyzes, as it's a one-time scan that remains anonymous.

Part of the impetus in doing this appears to come from a study Auto Trader commissioned called the CarCourting Report, which showed that "rational … emotional and even social implications play a major influence on Canadian car buyers". The poll showed that 71 per cent of respondents weren't "in love with their current ride".

Ian MacDonald, director of marketing at Trader Corporation, said that this could be a more accurate method of qualifying the consumer than a survey or questionnaire because it's based on data that is highly personalized to begin with.

That's not necessarily true if there is little data to go on. Some Facebook users may be fairly open about their lives on the site, but income isn't one of those things. MacDonald says budget is inferred by affluence parameters, like education, employment and locations where a user has been using the site's "check-in" feature. Those might help if a user was a highly active Facebook user to ensure greater accuracy. A sparsely-populated profile with little activity is likely to garner strange results, a fact MacDonald didn't disagree with.

"Like any algorithm, the greater the amount of data, the greater the accuracy," he says. "Users can share results with friends and get feedback from them as well, so we're looking to evaluate the overall usage and feedback of the app in the hopes that it can be a tool people can use to help them find the right vehicle."

The idea, of course, is to see if or how the app might influence purchase behaviour on the AutoTrader site, since results are directly tied to what's actually for sale in real-time. Car aficionados are likely to laugh at the autoLYZER, but those with only a passing knowledge are probably who AutoTrader is going after. In testing it with friends, I found results to straddle between surprising and bizarre in most cases. But then again, half were active users while the other half didn't use Facebook nearly as much.

globedrive@globeandmail.com

Interact with The Globe