FinTech · Consumer lending · United States

Ad spend matched to real applications.

Client name withheld under NDA 01 Digital analytics02 Data platform
The client

A US consumer finance company offering lease-to-own financing at the point of sale, aimed at shoppers with limited or damaged credit. It works through a large national network of retail partners, from furniture and mattress stores to tyre and wheel shops, and customers apply online, in store or over the phone. Marketing runs across search, social, programmatic and partner channels.

The challenge

Marketing was spending across several ad platforms with no way to connect that spend to the applications it produced. A large share of applications finished offline, in store or with an agent, so the ad platforms never saw them.

Tracking had grown piece by piece. Events were named differently across the funnel, browser-side tags lost data to ad blockers and iOS privacy changes, and each platform claimed its own version of the truth.

There was no marketing data warehouse. Ad platform data, web analytics and application records lived in separate systems, so every budget question turned into a manual spreadsheet exercise.

What we did
Server-side trackingMarketing data warehouseAttribution modellingConversion APIs

GA4 with server-side tagging

We designed a new event taxonomy around the lending funnel, from first ad click to submitted and approved application. GA4 was rebuilt against that plan and routed through a server-side GTM container on GCP, which gave the team privacy-compliant collection and far less data loss.

Marketing data warehouse in BigQuery

We built a BigQuery warehouse that brings together ad platform spend, GA4 behaviour and application data from the lending systems. Modelled tables join these sources on shared keys, so spend, sessions and applications sit side by side for the first time.

Custom multi-touch attribution

We built multi-touch attribution models in Python on top of the warehouse, covering a very large annual ad budget. The models credit each channel for its part in the journey and feed channel-level reporting the marketing team uses to plan spend.

Modelling untracked offline applications

Many offline applications had no digital trail. We developed logistic regression models that estimate how many of those applications came from marketing, giving the team a measured bridge between online activity and offline outcomes.

Conversion API integrations

We implemented server-side conversion feeds for Google Ads, Meta, Basis and StackAdapt. Each platform now receives application events it could not see before, so bidding algorithms optimise towards real applications.

SEO programme

We set up an SEO programme shaped around how lease-to-own shoppers research their options. Content and technical fixes were prioritised against the questions customers ask before they apply.

What changed
  • Marketing spend was measured against real applications, including offline ones, for the first time.
  • Budget moved away from channels that were not producing applications.
  • A bridge between offline applications and online activity now feeds measurement and ad platform bidding.
  • One warehouse replaced the scattered platform dashboards.
  • Tracking is resilient to ad blockers and browser privacy changes.
Stack
GA4Server-side GTMBigQueryGCPPythonGoogle AdsMetaBasisStackAdapt
About this case study

The client’s name and specifics are covered by an NDA that limits what we can publish. We have permission to discuss the work in more detail in private, so ask us on a call.

Book a call

Related work

Next step

Get your first fix in 30 minutes

Bring the question your team can’t answer fast enough. The engineer who would do the work looks at your setup with you and tells you the first thing we would fix, plus a straight answer on whether you need us at all. It costs you half an hour, and you keep the fix either way.

Book a free discovery call