How to Build an E-commerce Profit Forecasting System for Growth

Introduction

This blog delves into the process of developing an operating system for profit forecasting in e-commerce. It emphasizes the key steps involved in building a consistent and growing e-commerce business through qualitative planning, quantitative modeling, execution mapping, and data management.

Developing an Operating System for Profit Forecasting

• The speaker emphasizes that profit forecasting is not about predicting the future with pinpoint accuracy, but rather about creating a system for operating a business. This system involves planning and execution to build a consistent and growing e-commerce business.

• The goal of effective forecasting is to be close to the target, with a margin of plus or minus 10 percent. The speaker stresses the importance of being able to quickly identify and correct any inaccuracies in the forecast, rather than striving for absolute accuracy.

• The forecasting process involves four key steps: qualitative planning, quantitative modeling, building a map for execution, and implementing a system for data management.

• The initial step, qualitative planning, involves a thoughtful marketing exercise that considers non-linear revenue patterns and identifies peak moments throughout the business calendar.

• The speaker also mentions a product called the enterprise scaling guide available on their website, which provides tools and templates for implementing the forecasting system.

Key Points for Marketing Success

• Key promotional periods like Memorial Day, Back to School, and other cultural moments are crucial for marketing planning and product promotions.

• Advertising success is driven by outsized performance during these key moments, which impacts the effectiveness of machine learning tools on platforms like Facebook and Google in distributing ad dollars.

• Efficiency in media spend can be improved by driving unexpected outsized conversion rates, leading to increased value returns.

• Creating peak moments through promotion, product releases, and limited editions can leverage changes in conversion rates to drive unexpected returns, similar to the performance seen on days like Black Friday and Cyber Monday.

• Collaborative, broader marketing efforts and storytelling, rather than solely relying on Evergreen ad iteration, drive success in ad accounts by aligning with seasonal trends and generating peak moments.

• The forecasting process begins by designing a detailed 12-month marketing calendar that outlines every message, promotion, and revenue spikes to manage cash flow and drive success.

Understanding Customer Cohorts for Predictive Modeling

• Customer cohorts, both new and repeat, form the foundation for predictive modeling in e-commerce.

• Understanding the behavior of your business requires studying and analyzing cohorts, as they are the atomic unit of e-commerce.

• Distinguishing between total customers and active/healthy customers is crucial for predicting future revenue.

• A healthy customer cohort should show a growing active customer file, indicating an increase in future revenue.

• Building a predictive model involves understanding three key aspects: predicting existing customer future revenue, new customer revenue, and customer value over time.

• An advanced data science team uses regression analysis and back testing to create predictive models around customer cohorts.

• Back testing involves comparing the model’s predictions with the actual results to ensure accuracy and predictability.

Revenue Forecasting and Customer Acquisition Model

• The existing customer revenue forms the most predictable portion of the model, and accurate prediction of existing customer revenue is crucial for a strong revenue forecast.

• Businesses with a large portion of revenue coming from existing customers, up to 70%, need to be highly accurate in predicting this revenue for a strong model foundation.

• The composition of revenue from new customers versus existing customers affects the forecast’s predictability, with higher new customer revenue leading to more volatility in the forecast.

• Determining the new customer acquisition portion of the model starts with the business owner’s qualitative determination of an acceptable return on invested capital (ROIC) and the customer acquisition cost (CAC) they are willing to accept.

• A waterfall chart is used to illustrate the first-order average order value (AOV), current cost of delivery, and the CAC at various levels to aid businesses in determining their investment decisions.

• A spend and CAC relationship model is built using scatter plots to understand the relationship between spending and CAC and to determine the efficiency of spend at various CAC levels for acquiring new customers.

• The model helps to predict the additional spend and new customer acquisition based on the acceptable CAC, forming the secondary layer of the revenue model.

• The accuracy of the model is assessed by evaluating the correlation between the model-predicted CAC and the actual CAC, with a strong predictor indicating a robust model for predicting spending CAC.

• The retention model, desired CAC, and spend model form the inputs for building the growth map, which includes the marketing calendar, channel-specific media plans, spending, retention models, and a 24-month profit and loss forecast.

• The growth map is a detailed working document that incorporates all the key elements discussed, providing a comprehensive overview of the revenue forecasting and customer acquisition strategy.

• A visual representation of the forecast, including sales, cost of delivery, ad spend, operating expenses, and profit, helps businesses make informed decisions about their growth and investment strategies.

Optimizing Marketing Strategies with Data-Driven Approach

• The speaker discusses the process of setting marketing goals by extracting data from the status database and using it to build a model.

• They highlight the use of marketing calendar to plan key events and campaigns, including expected revenue from emails and social media advertisements.

• The media buyer then plans the expected spend for each campaign based on the marketing calendar moments and tracks actual spend daily to compare with expectations.

• The model also includes cohort input tab to create a Customer Acquisition Cost (CAC) model based on expected output at each spend level.

• The output of all the retention curves is then analyzed, leading to a detailed model encompassing channel specific plans and spend.

• The data from the model is imported into a data platform for tracking daily revenue, contribution margin, and ad spend, enabling quick adjustments based on performance.

• The daily tracking and adjustment process is integrated into a tool called ‘plot pivot profit,’ allowing for a daily understanding of performance relative to expectations.

• The speaker emphasizes the importance of the data-driven approach in managing growth, solving problems, and driving profit on a daily basis.

Conclusion

By implementing the discussed strategies, businesses can develop a robust profit forecasting system, optimize marketing strategies, and drive consistent growth in the e-commerce industry.

undefined