‍Putting Your Data
To Work.
Opportunity is in your data. But you need to ask the right questions to unlock it.

At Theory+Practice™, we are experts at helping large enterprises in finance, retail and consumer packaged goods put their data to work. By pairing behavioural economics with state-of-the-art AI technologies we help you truly understand your customers and build business strategies that drive results.
CASE STUDIES
Module Showcase
Our clients appreciate the unique question-first approach that we use to deploy our AI modules and the advantages that data insights bring to their businesses.
RETAIL AI SYSTEM
A simple white icon of a grocery cart
SMART 
PERSONALIZATION 
MODULE
Client: One of the largest supermarket chains in the United Kingdom with revenue over $17 Billion.

GOAL

Improve customer experience to drive repeat visits and loyalty while increasing basket size/value.

‍

ORIGINAL APPROACH

At the self checkout in stores, customer gets a pop up offering generic items to add to their cart.

OUR MODULES & RESULTS

    • The Product Recommendation Module seamlessly leverages customer preferences to make recommendations personally catered‍‍
    • Integrated and receiving data from 400 stores and live in 206 stores at self order point tills, providing customers with recommendations tailored to their session
    • The Product Recommendation Module is optimized for time of day, location, store type and store size to increase efficiencies

    Outcomes

    • Average store conversion rate of recommendations ranges from 7% to 10% and varies by time of day. Some stores experienced a conversion rate of 12%
    • Average value contributed to baskets by the Product Recommendation Module ranges from 2% to 3.5% depending on the store, with some baskets attributing up to 40% of the value from recommendations

    ‍

    ‍

CPG AI SYSTEM
Simple white icon of a truck used to ship products
DEMAND 
FORECASTING
MODULE
Client: Large CPG in specialty fresh products and bakery, with revenue over $1 Billion.

GOAL

Get a highly accurate demand forecasting tool to provide an ongoing and detailed understanding of the impact of changing prices and other marketing initiatives on sales and product demand.


‍

ORIGINAL APPROACH

Sales specialists used year old 3rd party price elasticity data to intuitively predict upcoming demand. Previous sales volume was used as the primary demand indicator.

Marketing attribution was executed based on educated guesses and the limited available data.

Compiling data manually in spreadsheets accounted for 10% of each sales expert's time, and their demand prediction accuracy was typically 75%.

OUR MODULES & RESULTS

  • Our Demand Forecasting Module for CPG includes:

    • The ability to learn trends
    • The Predictive Marketing Attribution Module, which combines advertising and sales data to provide in-depth Marketing Attribution analysis available on-demand for a specific region, product, time-span, or campaign

    Outcomes

    • ‍Aggregate model accuracy of 96.5% for demand prediction
    • Aggregate Weighted Mean Average Prediction Error (wMAPE) is less than 3%‍
    • Client can now perform scenario planning and what-if analysis on demand. For example, what are the sales in the next prediction window, given a level of price increase?‍
    • The new depth of understanding and on-demand simulations empowers the business to always create the best strategy
    • The Demand Forecasting Module also enabled the business to optimize merchandising and pricing with forward-looking lens through simulations

    ‍

    ‍

FINANCE AI SYSTEM
A simple white icon of a bank
CAMPAIGN
MODULE
Client: Leading Credit Union with over $5 billion in total assets under administration.

GOAL

Increase conversion rate and revenues by communicating personalized offers and products to the right members at times they are most likely to commit.

‍

ORIGINAL APPROACH

For a typical campaign, the credit union would call groups of members that have had terms deposits in the past or call as many members as possible.

The conversion rate is 3% or less on a typical campaign for customers without term deposits.

‍

OUR MODULES & RESULTS

  • Our Campaign Module for Finance generates campaign lists that were integrated into the campaign workflow on-demand and includes:

    • Lists of members with the highest likelihood of acquiring several different financial products such as term deposits, loans, mortgages, wealth management, etc
    • Insights from the Behavioural Segmentation Module that identifies micro-segments
    • Customized and personalized talking points for call center, branch staff, and digital campaign that increase members engagement and conversions
    • Identifies members most at risk of dropping out and appropriate interventions

    Term deposit campaign outcomes

    • Campaign 1 resulted in a 27% average conversion rate generating $30.2M in term deposits in 8 weeks‍

           - 54% increase in the number of term deposits

           - 50.5% increase in the amount committed to            term deposits

    • Campaign 2 generated $52M in 4 weeks

      ‍
FINANCE AI SYSTEM
A simple illustration of a hand holding a house with a percentage sign in it, to reflect mortgages
ORIGINATION & MATCHING MODULE
Client: Leading organization selling mortgage and non-mortgage lending products.

GOAL

Increase its lead conversion rate through dynamically matching the best loan officer to each lead, to increase close rates, efficiency, and revenue growth.

‍

ORIGINAL APPROACH

Purchases costly Lead lists each month for call center agents to use for cold calling – producing very low conversion rates.
‍
Lead lists are divided among five clusters of Loan Officers randomly.

OUR MODULES & RESULTS

  • Our Origination and Matching Module consists of:

    • Origination Module that scores Leads based on their likelihood to originate/get funded in general and the likelihood to be funded by competitors
    • Matching module that uses a robust neural network model to match the Leads to the best available Loan Officer
    • Segmentation Module that creates a deep understanding of Leads and Loan Officers

    Outcomes

    • Propensity to originate/fund models allowed the mortgage lender to prioritize incoming Leads based on desired KPI’s
    • The Dynamic Lead to Loan Officer Matching Module predicted the lead conversion with 87% accuracy and results in a 66% increase from the historical baseline
    • The deep segmentation insights combined with the origination and matching outcomes enabled the mortgage lender to focus its initial screening effort on the Leads that are more likely to be funded

    ‍

    ‍

    ‍

Ready to put your data to work?
Get in touch with our elite team of data scientists and behavioral economists and discover the insights you need to supercharge your business.
Contact US
About us
ExpertiseWorkTeam
ContactPrivacy
Sign up for the Theory+Practice newsletter for exclusive access to case studies and the latest in industry news and advancements.  
Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.
All Rights Reserved. 2023.