“Eliot, we are worried that our advertising activities are not profitable… can we tell us how much our maximum cost per lead should be in order for us to be profitable?”. This question came from Jérôme Stefanski (CEO), Pauline Jobelot (CMO) and Stéphanie Fay (Growth Marketing Manager) from Little Guest. And this is how this case started.
With lead generation clients, it’s hard to tell how much the cost per lead should be without the proper tracking in place. Sure, we can always ask our clients what their average conversion rate (from leads to sales) and their average conversion value is to have a broad idea of the return on investment of our campaigns. But this still remains a very high level estimation which could end up being very far from the reality of our campaigns. In addition, this estimation does not allow us to invest more efficiently either since we still do not know who the “good” and the “poor” leads are. Therefore, how can we get closer to the business of lead generation clients like Little Guest? This case will be focussing on solving this question.
The case will be divided into 2 parts. The first one will explain the business logic and high-level implementation flow, while the second will focus on the actual technical implementation. This is the first part of the case.
Who is Little Guest ?
Little Guest is a high-end family-oriented hotel collection company. The company offers a wide variety of 4 and 5 stars hotels all over the world, with a particular attention to the quality of services offered to parents and their children. You can discover more about Little Guest on their website.
Today, there are currently two ways to book a stay on Little Guest’s website. For some hotels, it’s possible to make the entire reservation process online. For others, the customer has to send a price-offer request, in which case he is recontacted later on by one of Little Guest’s travel designers to complete the reservation. This double flow makes it more complex to compute the profitability of our advertising campaigns. While the direct sales are easy to track, which value should we attribute to the people sending price-offer requests?
To be able to compute the quality of the leads and the volume of sales generated by them, we needed to create some kind of link between our campaigns and the client CRM. Since the leads are not purchasing travels directly, it’s indeed necessary to find a way to remember where the leads were coming from so that we can attribute their future booking to the right channel. To do so, multiple elements had to be captured and sent to the CRM at the moment of the price-offer request. We chose to send the campaign names and the traffic sources & mediums. With this information, it was already possible from the CRM side to know exactly how much bookings and revenue our campaigns generated.
This was good, but still not very handy. We wanted to have this information directly available in our web analytics and advertising platforms in order to have all the results centralized. To achieve this, we needed to use an unique identifier so that the various platforms could recognize the users & clicks. For Google Analytics, we used the user ID to recognize the users, and for Google ads and Meta, we used the gclid and the fbclid to recognize the clicks. From the moment these elements were captured, it was finally possible to send back the “offline” booking results back to Google Analytics (through the measurement protocol), Google Ads (through conversion import), and Meta (through offline event data import).
Once everything was implemented correctly, we could compute the total return on investment of our campaigns for both the online and “offline” sales. Interestingly, at that moment, we realized that there was really nothing to worry about as our ROI in Google Analytics was actually about 5 times higher than the minimum required for Little Guest to be profitable.
Measuring is good, but activating is even better. We did not just want to measure our profitability, we wanted to increase it. To achieve this, we started using the offline sales data imported in Google Ads & Meta as conversion optimization goals for the algorithm instead of simply the price-offer requests. By doing so, we were basically telling the algorithm who were the qualitative and non-qualitative leads so that it could optimize our campaigns even more effectively than before.
There were nonetheless two issues with this strategy. On the one hand, not all our campaigns were receiving enough conversions data (both online & offline sales) to be optimized efficiently. On the other hand, travel designers sometimes had a hard time keeping up with the volume of leads entering during some very busy periods, meaning that some very qualitative leads were not purchasing their stays with Little Guest by lack of follow-up of their requests. In that situation, the non-converting although very qualitative leads were attributed the exact same value as the extremely poor lead - 0€ - making it hard for Google Ads & Meta algorithm to understand who actually were the interesting leads. Because of these issues, we put in place an automated lead scoring method (based on the data collected on the user & his request), allowing us to still attribute a value to non-converting leads based on our estimated chance of future booking. By doing so, we had increased the amount of conversions sent & collected by each campaign, while differentiating non-converting good & poor quality leads altogether.
In this case, we showed how we were able to overcome the issue of measuring and activating leads requests value. By capturing the traffic source, medium and campaigns, as well as the unique identifiers (for the users or the clicks) at the moment of the price-offer request, we were indeed able to identify the source of each sale and attribute the value back to the correct traffic source & campaign. Doing so also allowed us to bid more efficiently by using that conversion value information to inform Google Ads & Meta algorithms on the quality of the leads, and hence push them to invest more on the more qualitative leads and less on the poor ones. This is a perfect example of how a great client-agency collaboration & alignment can lead to better business results.