There are a variety of business scenarios where it is desirable to simulate the results of a controlled test, but no robust control cell exists. For example, marketing campaigns and other communications are often sent to customers based on their profile, without “holding out” a subset of customers to study later; web and interactive marketing is often done based on customer profile and behavior without creating a control cell; and changes in product and customer service are often done to entire populations simultaneously.

To combat these issues, the use of “synthetic” control groups has recently been adopted in several contexts, with generally positive results. Several firms have recently adopted the synthetic control approach to extract actionable insights from historical actions without a robust control sample. Two examples include:

1. Fortune 100 Financial Services Firm:

At a Fortune Global 100 financial services firm, the marketing group have used a multichannel communication strategy to market to their customers and prospects. The group wanted to understand the relative impact of these various marketing efforts on their collective success metrics and KPIs.

Through the use of synthetic control groups, they were able to quantify the financial impact of each marketing campaign by segment, and measure the change in scorecard metrics such as capture rate, retention, and reactivation. This allowed marketing leaders to reinvest a substantial share of their marketing investment into high impact activities, adding millions to the company’s top and bottom lines.


2. Fortune 50 Health Care Firm:

Executives at a Fortune 50 health care firm wanted to measure which of their Care Management programs had maximal impact for specific member segments. A synthetic control approach enabled them to retroactively evaluate the impacts of various programs across member segments. The use of a synthetic control group also helped validate the financial and medical value of each program, which in turn drove increased sales and retention.  Such an exercise would have been prohibitively onerous and expensive had it required actual in-market controlled testing.

In order to validate the statistical efficacy of this method, we have tested the results using actual in-market campaign test and control groups. This testing shows a correlation of 89% between actual control and synthetic control results.



Synthetic control groups represent an actionable, flexible approach to approximating the results of a controlled test where no actual control group exists, or the control group is too small to draw desired conclusions (e.g. segmented results). It is applicable in a broad range of business scenarios, and when implemented with appropriate care can yield results approximating those of a true control group.

For information on FischerJordan’s solutions for creating synthetic control groups or marketing spend optimization, let’s talk!  Talk to Us