Many Industries — Stacked Solutions
If you can judge a person by the company they keep, you can judge a company by the people that keep coming back. We’re in the business of fixing complex data problems, making sophisticated platforms work in a personal way to produce better outcomes. Here is some of our recent work:
CLIENT SITUATION > Data failing to meet business development needs
PROBLEM > A medical insurance provider wanted to increase lead acquisition for their business and was struggling to evaluate lead quality and price in a real time bidding situation.
STACKED SOLUTION > Using an AWS serverless application model, RDS, AWS Lambda and Sagemaker, we created a machine learning model and API that ingested the factors for each lead to determine whether to acquire it, examining data points including location, lifetime value, and other factors contributing to lead close. We integrated the findings with Salesforce to supply feedback to the model to support continuous learning.
RESULT > As a result of the new stacked approach, the client increased lead volume by 480% and lowered the cost-per-lead by 50%, all while maintaining a steady conversion rate of incoming leads.
CLIENT SITUATION > Fast growth direct to consumer startup needed data stack and self serve access to critical business metrics
PROBLEM > Honeylove, a venture-backed direct-to-consumer fashion brand, needed a robust data and analytics platform to support their rapid growth. They had exellent baseline analytics in place but had not yet integrated their operational, marketing, and digital data to unify their understanding of customer, product, and marketing activities.
STACKED SOLUTION > Our prior experience with ecommerce brands and analytics methods enabled us to drive the design of the initial dashboards, the success of which led to the development of a full data warehouse and the implementation of a BI reporting tool.
For the data warehouse development, we took a holistic approach, understanding data connectors and all the elements that needed to be pulled together – from merchandising and logistics to customer service and affiliates. We fulfilled this through Stitch, Fivetran and custom connectors.
Once the data became robust in Honeylove’s Big Query warehouse and Data Studio, end users soon demanded richer insights and self-serve access, which clarified the need for a new BI platform and the implementation of Looker.
Throughout the engagement, we provided expertise in other specialties, including approaches around GA4, server side rendering, cookie compliance, employing OneTrust, sizing tool usage and style quizzes.
RESULT > Looker is now fully deployed and users are engaged in the data. The structure and systems are in place to readily and easily add new data assets including the ingestion of offline data, as Honeylove is now selling offline in retail stores across the U.S. Additionally, business users are relying on the dashboards to better manage and understand discounts, returns, influencers, inventory management, garment fit, orders, shipping SLAs, customer service experience and loyalty programs.
CLIENT SITUATION > Needed an outside strategic perspective to uncover greater conversion value from data platform
PROBLEM > A large higher education company was struggling to improve conversion rates from their paid media campaigns as they saw their costs rising. The organization had multiple forms that fit different prospective students' situations. Staffing of the call center came into play, as well, since the conversion forms required different call center staff to respond to the form completion.
STACKED SOLUTION > We worked with the company to understand the organizational constraints and the value of each conversion form. We analyzed the data points known at the time of form selection (time of day, browser, previous visits to the site, medium/source, etc) to determine which of those points made a difference to completing each form type. We built a machine learning model to predict completion of each form, and an API that told the website which form to send to a user, maximizing the higher value conversion form within the call center staffing constraints.
This change allowed the website to offer one conversion button instead of two and made the best choice for the user based on the machine learning model.
RESULT > These efforts resulted in a 25% increase in conversions and a 7% increase in application starts.