Autobytel has reinvented itself a few times since becoming a pioneer of providing sales leads to auto makers and dealers.

The 16-year-old company is transforming itself, again. “It's back to the future,” CEO Jeff Coats tells me.

The Irvine, CA-based firm's new strategy is to generate more of its own leads, and rely less on aggregating them from various third-party sources.

Home-grown leads offer higher margins and better close rates, Coats says.

The latter leads to happier dealers who feel they got their money's worth from a purchased lead. Sometimes they don't feel that way. Not all leads are created equal.

“By concentrating on internally generated lead requests,” Autobytel will “continue to aggressively move away from a purchase-request aggregation model,” Coats says.

The firm expects to spawn 65% of its own leads this year, he says. “Previously, we were generating less than 5% of the leads we sold. It was during a proliferation of lead generation. We became like everyone else. In our early years, 100% of our leads came from us.”

The quality of online sales leads can put dealers and lead providers at odds.

Dealers will complain a lead stinks, while providers beef about dealerships mishandling perfectly good online prospects sent their way.

Either way, lead quality indeed varies, depending on the source.

For instance, good leads come from consumers who are actively shopping for cars and submit purchase requests through automotive websites.

Lousy leads come from Internet users who land on non-automotive sites and submit leads as a condition to winning contest prizes.

A dealership Internet manager tells of following up on a lead only to have the person on the other end of the phone say, “I'm not interested in buying a car; I'm interested in winning a trip to Las Vegas.”

Search-engine marketing tactics can aid and abet the creation of bad sales prospects by steering people to those “submit-a-lead/win-a-trip” sites.

“A company consisting of two guys and a dog in a garage could generate cheap leads in the age of search,” Coats says.

Not surprisingly, problems occur when dead-end leads like that are sold to aggregators, who turn around and sell them to dealers, who end up miffed.

To avoid such dealer dissatisfaction, Autobytel is shifting from buying fewer leads to cultivating more of its own. Helping the cause is the acquisition of Autoropolis and Cyber Ventures, two firms known for generating decent leads.

Dealers display different preferences when it comes to leads, says Stephen Lind, Autobytel's executive vice president-corporate development.

“Some dealers like a wider regional leads,” while others want leads from local prospects only, he says. “Dealers can buy leads by car makes and models, and a lot of them do that.”

Proper lead handling depends on consistent follow up by the dealership, Lind says. “It's not always a one-call close. Timely follow ups make a difference. We spend a lot of time in the field emphasizing that.”

Autobytel expects to deliver 3.8 million leads in 2011. By switching strategies, “we're “going back to our roots,” Coats says.

Well, not entirely.

The company began in 1995 by faxing leads to dealers. “Back then, most dealerships had leased computers and most didn't have email,” says Coats, who has been associated with the firm since 1996 and CEO since 2008.

In 1998, Autobytel (the “tel” standing for “telephone”) made the big leap to the Internet.

The firm recently revamped its website to make it livelier and easier to use.

Focus group and surveys indicated consumers felt sites in Autobytel's category were not “customizable” and often assumed all visitors were alike and their needs the same, the company says of its market research.

Consumers said they visit such sites for different reasons and at varied stages in the car-purchase process. They indicated the experience could intimidate them. They expressed a desire for “fun” while car researching and shopping online.

The new website was designed to provide just that, Coats says.