Web analytics uses data that are unique with respect to the quantity and detail of information that is available. What is consistent with web business analytics and analytics for business activities not on the web, is their purpose: to improve business performance.
Business performance can be defined in many ways depending on the context, but maximizing economic profit is the ultimate objective. Discussed here are nine key business analytics that a digital business should complete to maximize its operating performance regardless of how it is measured. These analytics can leverage the detailed data available on web activity as well as other data on a company that is available.
Business Tactics and Analytics
Web-based data can capture unprecedented details about how customers make their purchase decision and how much they value the product or service. Business strategies and tactics that can be supported by analytics using web data are sales offers, product bundling, renewal offers, retention touch points, cross-sell offers and up-sell offers. For these activities, common analytics are segmentation using clustering approaches, propensity models using discrete choice econometrics, retention analysis using survival analysis and testing using statistically valid samples for control groups.
A table of these business operations and supportive analytics
1 – Customer Acquisition
Achieving sales to new customers is one of the most challenging yet critical activities a business performs. Web data provide insights into how this process occurs and what factors influence the purchase decision. Applying these data in an analytics framework that can isolate the influence of individual factors can help a company avoid costly mistakes, such as under-pricing or overpricing a product.
Discrete choice models, called thus due to the binary outcome of the customer decision to accept or not to accept an offer to buy a product or service, is a common tool for measuring the influence of each of these factors. Promotional offers are a tactic often used to persuade customers to buy a product. Understanding the effect that price has on a consumer’s choice is enormously valuable and often achievable using discrete choice models.
2 – Renewal and Retention Offers
Once a customer has been acquired by an organization, retaining that customer on a recurring service, such as a subscription, or keeping them loyal to a brand for a repeat purchase, is a critical business activity and a complex analytical topic. Survival analysis is an econometric tool developed in the healthcare field, and it is effective in measuring the influence of separate factors on customer retention, including changes in the price of the product or service at the end of a promotional period. Discrete choice can also be used to analyze repeat-purchase decisions.
3 – Product Bundling
Companies often can sell more of two products bundled together than they would sell of them separately. The key to successful bundling is to understand the relative value that customers place on the components of a bundle and what product features are most important to specific customer segments. Web data is very helpful in measuring the effect of bundle components or product features on purchases. A common analytical tool for addressing these questions is hedonic price modeling.
4 – Customer Segmentation
Grouping customers into segments with similar preferences and price elasticity enables businesses to design product bundles and pricing strategies that maximize operating margins and economic profit. Using cluster analysis with web data included along with other data, companies can create segments within their customer group. K-means clustering starts with the analyst specifying the number of clusters to be created. Hierarchical clustering is much more computationally intensive, but it is effective in determining the optimal number of clusters.
5 – Price Elasticity Measurement
Estimating price elasticity, the percentage change in quantity purchased due to a percent change in price, is typically the most important analysis that a company can undertake. There are many types of econometric modeling that can support this objective, but regressions using panel data is a common approach. For businesses with subscribers, price elasticity typically declines with the customer’s tenure with the product. Understanding the rate at which price elasticity changes over time is helpful in maximizing the lifetime value and economic profit from a customer.
6 – Up-sell & Cross-sell Offers
Expanding a relationship with a customer to include more products or services (cross-selling,) or enhancing a customer’s existing product or service to include more features (up-selling) is an effective tactic for increasing revenue and retention. Designing sales offers that maximize the economic profit of these tactics can be supported with discrete choice models to identify offer characteristics that influence their effectiveness. Higher prices will tend to have lower acceptance rates and vice versa. Analytics can help companies balance the quantity versus quality tradeoff.
7 – Forecasting
Estimating the future level of sales reduces waste from overproduction or lost sales from underproduction. Perishable inventory businesses, such as airlines, hotels and advertising companies, need forecasts of sales over time to optimize revenue per unit of inventory. Data from digital operations are superbly suited for forecasting models. Time series econometrics is the field of analysis most often used for forecasting.
8 – Customer Profitability
Understanding which customers are providing your economic profit is critical to maximizing business performance. Too often in digital businesses, top-line revenue is the focus of the organization and what is rewarded. Many businesses spend inordinate amounts of time and resources growing business lines that return very little operating margin to the organization and do not cover the opportunity cost of the capital required for those activities. Lifetime value analysis that includes direct, variable costs and revenue along with a forecasted probability of future sales is an excellent tool for scoring customers and understanding profitability.
9 – Test and Learn
Predictive analytics are only as valuable as the effectiveness of the business actions they support. Complementing applied analytics with robust testing provides data on the accuracy of the models used to make decisions. For instance, using a statistically valid sample of customers to test a pricing action can validate the price elasticity measured by an econometric model estimated using data from prior sales activity.
Conclusions
Analytics using data from digital business operations can provide a competitive advantage for a company. As these types of analytics proliferate, a company’s decisions regarding tools will no longer be whether to use them but to what degree and in what manner.
The nine types of analytics discussed briefly are not the only important analytics companies should complete, but they are some of the most vital.
Leave a Reply