Sign in with Google
As Chief Customer and Marketing Officer, Ben Carter is responsible for global marketing and driving customer centricity across Carwow. He and marketing analyst Roman Petrochenkov implement new marketing measurement strategies for business growth.
The automotive industry is in constant motion, driven by technological advancements and evolving consumer behaviours. The rise of online car marketplaces has transformed how people research and purchase vehicles, creating even more touchpoints in a typical customer journey.
In this dynamic landscape, Carwow’s recent evolution from a single-product platform to a multi-product, end-to-end car changing destination, offering new and used car buying, selling and leasing options, has significantly increased the complexity of our user journeys. Coupled with changing data privacy regulations and a proliferation of marketing channels, it’s more difficult than ever for us to measure how people move from research to purchase.
We knew we needed to get a more complete picture of customer behaviour to help us drive sustainable business growth. That’s why we tapped into an innovative solution that uses first-party data to future-proof our measurement infrastructure and we were amazed by the results we found.
Navigating complexity and change
Over the past three years, we’ve updated our measurements practices significantly, and we knew that these small changes directly impacted our profitability. For example, we moved away from last click and started using value-based bidding, and shifted our optimisation focus from customer sign-ups to revenue-generating events — like leads from enquiries for new car sales.
But we were still relying on offline conversion import (OCI) to feed conversion data. OCI is a Google Ads feature that allows you to import conversion data from offline sources, like phone calls, back into your account. This is essentially letting you measure how online ad clicks translate into offline actions.
It does this using a unique identifier like a Google Click ID stored in your CRM system. These Click IDs help identify the campaign and other attributes of the ‘click’ on an ad for tracking and campaign attribution purposes. E.g.: customer X saw ad campaign Y, and clicked onto the site.
While effective, with OCI you can’t measure certain events. For example, you usually lose the ability to measure conversion events like cross-device conversions, engaged-view conversions or conversions where the Click IDs have been removed by browser restrictions.
Shifting gears with enhanced conversions
Then we found out about Enhanced Conversions for Leads (ECL).
It is an ads feature designed to improve the accuracy of conversion measurement — exactly what we needed. It would allow us to focus our future bidding on the channels and campaigns that were working best for us.
It works by using first-party data — e.g. from website lead forms — to improve the accuracy of conversion measurement and bidding performance, especially for offline sales. Here’s how it works:
We were excited about using Enhanced Conversions for Leads because:
- We could see the value of using our first-party data. It comes directly from our own interactions with customers, provides more reliable insights, enables better personalisation, and strengthens relationships — without relying on third-party sources.
- It would help us understand more complex customer journeys that are most likely to be impacted by the loss of Click IDs.
In fact, we learned that advertisers who upgrade from standard offline conversion import to enhanced conversions on average see 8% more conversions on Search and 22% more on YouTube.1
We wanted to validate this claim for ourselves so we developed a three-phase testing model. Warning, this is the technical part:
1. We tested Click ID vs Click ID and user provided data
Essentially in this phase we added user provided data (email addresses) to enhance Google conversions with Click IDs. We tested this against just sending Click IDs, with the latter limiting the ability to connect actions driven by other ad interactions, therefore proving that adding the additional data increased conversions slightly.
2. Click ID vs user provided data
In the second phase we only sent user provided data (email addresses) to see how good or bad measurement would be vs Click IDs. We were able to reconstruct approximately 90% of conversions even when we didn’t have enough click information.
3. Click ID vs Click ID and user provided data for all conversion data
Finally, we tested Enhanced Conversions for Leads with all best practices — sending Click IDs when available, as well as user provided data for all conversions, including those where we did not have a Click ID at all.
We needed to be thorough about how we adopted ECL, and validating the performance in this way gave us the confidence at every stage to progress to the next level of implementation.
Accelerating performance results
From this phased testing approach and with ECL technology, we were finally able to attribute credit to upper-funnel channels more accurately — something that wasn’t fully possible with a click-based model alone. While we always understood their importance, previous measurement methods didn’t capture their true impact.
This is because for mid- and upper-funnel campaigns, consumers don’t necessarily click directly on an ad before visiting a website. Instead they may, for example, watch an ad, continue watching the YouTube video, and then visit the website.
A click-based attribution model would not attribute any credit to the ad in this example. Understanding this was crucial for us, as the car buying journey is often lengthy and complex and we identified the need to be present throughout to support our customers on their journeys.
More importantly, Enhanced Conversions for Leads allowed us to capture and attribute value from customer journeys that were previously lost due to the growing tracking limitations. We were able to stitch the journey for those users together and find the campaigns that had driven those users to us.
This allowed us to actually prove that mid-funnel channels, such as YouTube, had a far bigger impact than previously observed in our measurement. In fact, we saw a 7% increase in attributed conversions compared to Click IDs alone. This means Google Ads was able to find up to 7% of additional conversions that we couldn’t upload clicks for.
We saw an increase across all campaign types, with channels that drive more direct click interactions such as Search up to 5%, and upper funnel formats such as Demand Gen up 24%.
These insights empowered us to optimise our marketing strategies, allocate budgets more effectively — by investing more in our mid- and Demand Gen activity for example.
Fuel for future change: The measurement triangle
While we’ve made significant strides in attribution, we recognise that this is just one piece of the puzzle. We’re now focusing on incrementality testing, refining our marketing mix modelling (MMM), and looking at solutions like Meridian.
Our goal is to improve our “measurement triangle” encompassing attribution, MMM, and incrementality. We’ve made great progress on attribution, are in the early stages of MMM, and are just beginning our journey with incrementality.
As a growing business, we need to adapt and evolve quickly, particularly as our business becomes more complex. We have to ensure that we are making the best data-driven decisions to drive sustainable growth.
