for a Large Caregiver Marketplace
for a Large Caregiver Marketplace
The primary challenge was to develop a comprehensive attribution model that could accurately measure the impact of multiple marketing channels, both online and offline. Traditional attribution models, which often assign full credit to a single touchpoint, were insufficient for the client’s needs. They required a fractional model that would distribute credit among various touchpoints and channels, ensuring a more accurate representation of the customer journey.
Stellans developed a tailored forecasting model to meet the client’s needs. The solution included:
Conversion events were set up to fire into GA4 (Google Analytics 4), covering three relevant types: lead, subscriber, and re-upgrade. UTM parameters were used to connect end users with the digital campaigns that brought them to the site.
An online survey was used to gather data on customer behavior and preferences. This included questions about how customers heard about the service, allowing the model to incorporate offline sources like TV ads and direct mail.
Weights were assigned based on the type and number of inputs (digital, offline, survey). Time decay coefficients were introduced to adjust the weight of touchpoints based on the time elapsed between the touchpoint and the final conversion.
Based on consultations with the client, the model was built entirely in SQL, leveraging Fivetran’s built-in dbt capability. The model assigns weights to each input signal, normalizing them to ensure that the resulting total weights add up to 1. This prevents double-counting and allows for a fair distribution of credit among various touchpoints.
The client used a multi-agency approach for various digital campaigns, including streaming audio, podcasts, and non-linear TV (OTT). Each agency tracked user behavior and conversions using internal attribution platforms. Data from these platforms were integrated into the data warehouse via multiple Fivetran connectors, including TheTradeDesk, GoogleSheets, email with CSV attachments, and SFTP
As users became leads by completing the initial onboarding process, digital footprints were tracked using UTM parameters. Survey responses were mapped to specific marketing channels, and offline inputs were integrated via agency feeds. The fractional attribution model then processed these inputs, providing a comprehensive view of how different marketing channels contributed to lead conversions.
The premium conversion process involved additional steps, including a premium self-attribution survey. The model used time decay coefficients to redistribute weights based on the time taken to convert from lead to premium, ensuring accurate attribution of marketing efforts.
The re-upgrade attribution process was standalone, tracking conversions from dormant to active premium users. The model took survey data and digital footprints into account, ensuring accurate attribution for re-upgrade conversions.
Hello,
Thank you for your interest in Stellan’s Data Services. After you fill out this form, one of our team members will get in touch with you at the requested time to discuss your tech initiative.
We can’t wait to deliver the best results to you!
Hello,
Thank you for your interest in Stellan’s Data Services. After you fill out this form, one of our team members will get in touch with you at the requested time to discuss your tech initiative.
We can’t wait to deliver the best results to you!
Hello,
Thank you for your interest in Stellan’s Data Services. After you fill out this form, one of our team members will get in touch with you at the requested time to discuss your tech initiative.
We can’t wait to deliver the best results to you!
For contacting us!
We'll contact you as soon as
possible.
The report is on its way and should arrive at the email address you provided shortly.
