LeLynx Restores Conversion Rates & Cleans Up Traffic With DataDome

The challenge: Unreliable KPIs & stolen data

Léaud Le Bacq is Web Project Manager at LeLynx.fr. He and his team started having problems with traffic and bad bots at the beginning of 2017.

“We began noticing irregular, unexplained drops in our conversion rates,” says Léaud. “When we dug a little, we saw that certain user profiles kept coming up over and over, and their behavior didn’t line up with user targets or expected segments. Over time, we realized that they might be caused by insurers creating fake profiles on our site to extract prices for competitive intelligence purposes.”

LeLynx.fr lets users obtain personalized estimates based on criteria that can be very detailed, and puts them in contact with insurance companies offering policies at competitive rates.

If bots fill out the forms using every possible combination of options, without generating any contacts in the end, it considerably skews conversion rates. Automated activity of this kind can consume huge amounts of bandwidth as well.

“Giving bots free rein also seemed dangerous to us,” adds Léaud. “For example, a bad actor could create bots to establish contacts with fake profiles, creating fake leads for the insurance companies that we wouldn’t have been able to charge for. It hadn’t happened yet, but we knew that if it did, it would cause far more serious problems than the difficulties interpreting our KPIs.”

The solution: DataDome runs autonomously to detect & block scraper bots

At first, Léaud and his team tried to solve the problem by implementing database rules, such as limiting the number of sessions allowed per IP address. But the bot operators quickly found ways to get around the rules.

“We finally decided to look outside the company for someone with the bot detection expertise we lacked,” he says. “We started with a benchmark of three or four different solutions, including DataDome, which we ended up choosing.”

At first, since the team already had a CAPTCHA solution, they planned to use DataDome only for detection, without activating the protection.

“The DataDome solution provided us with the detection data, telling us, ‘This IP is a robot,’ but we blocked the IP addresses ourselves and decided how long they would stay blocked before being reauthorized, for testing purposes,” he explains.

This method worked well in the beginning. But in early 2020, during the first national lockdown in response to the coronavirus pandemic, LeLynx was targeted with massive bot attacks. The internal solution couldn’t handle them, and the security system was bypassed.

As a result, Léaud and his team took the time to reinstall DataDome and activate the protection, with help from DataDome’s onboarding team.

The results: Conversion rates restored & prices protected

LeLynx.fr doesn’t get attacked anymore, now that the DataDome protection is activated. That’s fairly common: when bot operators discover a protection system they can’t get around, the attacks are no longer financially viable, so they eventually give up.

For Léaud, it means that his KPIs and traffic volume are stable and reliable again.

“For us, that’s the main criteria for success,” he says. “No more surprises, no more attacks, and no more sudden spikes in volume. It’s particularly reassuring given that other competitors in the market have been hit with serious attacks recently.”

The team had initially implemented DataDome only for their most critical service, car insurance. Now they’ve decided to expand protection to cover their entire product selection.

“We’ve configured the solution well. On the dashboard, each product is defined as its own endpoint. It makes it very easy for us to see the level of bot activity in car insurance, motorcycle insurance, health insurance, etc.” Léaud explains.

In practice, however, he doesn’t spend much time on the dashboard. DataDome was designed to function in autopilot mode, and doesn’t require daily user intervention.

“I just look at the DataDome graphs from time to time, and I only dig deeper if something about the numbers catches my eye,” he concludes