Accelerating Your Best Ideas to Market
By John Cripps, Ph.D.
Frequently, we find ourselves in situations in which we have a new product or service that we would like to bring to market, yet many key questions remain unanswered, such as:
What will be the impact on trial and repeat sales rates in a competitive marketplace? What is the optimal product or service bundle configuration?- Which features are most appealing and in what combinations?
- How should the product or service be priced?
- How should the product be branded or cobranded?
How many and which bundles should be offered in the product or service line? What is the trade-off between additional products and incremental profitability?We've been using an approach we call Test AccelerationTM to address the drawbacks of traditional in-market testing — time consumption, expense, organizational politics, and loss of competitive secrecy.
 |
Test Acceleration, either as a precursor to a direct rollout of new products/services or to in-market testing of a much-refined concept, uses:
the Web as a consumer lab, discrete choice modeling to quantify consumer decision making, and constrained optimization.The Web is an extremely efficient way to survey customers and prospects and present them with rich descriptions of new products or services. It allows large survey samples to be collected quickly from targeted geographies and customer segments at a very low cost. Overall, we've found we can reduce testing cycle times, speed innovations to market, drastically reduce total testing costs, and significantly improve success rates through the testing of many more variables than would be possible otherwise.
The approach is based on discrete choice modeling, which was first introduced in the 1970s by Daniel McFadden (University of California, Berkeley), who recently won a Nobel Prize for his work. The technique presents consumers, through computer-assisted surveys, with hypothetical choices that force them to make tradeoffs between elements or attributes of an offer (credit card annual percentage rate, card color, credit line, rewards program, etc.) and reveal inherent preferences. The choice data collected is analyzed using statistical methods that allow a simulation of the impact of offering any possible combination of features/attributes — providing insight into the appeal of versions far beyond the specific set of offers that were explicitly shown in the survey.
But it gets even better. This type of quantitative choice model more than simulates what happens if we make an offer to a consumer segment or market.
|
|
|
|
||||||||
|
E-mail invitations are sent to respondents inviting them to take part
Quota sampling and weighting used to ensure relevant customer set is targeted without skew Lottery incentives are used to motivate respondent participation |
Respondents answer a series of hypothetical choice questions about:
• Which marketing creative they prefer, if any • Which offer they would choose, if any Offer and creative design combinations are varied systematically in the choice questions |
Answers to choice questions are used to build "choice models," which predict:
• Response to marketing creative • Response to offer design and price Offer and creative costs are combined with choice models to simulate response to all possible offer and creative combinations |
Optimization algorithms are used to search for best offer and creative combinations
Diagnostic impact is measured for each offer element (e.g., price elasticity) and creative element (e.g., e-mail subject line) Best offer and creative element combinations are then either: • Tested in-market to validate and refine • Rolled out to the market directly |
We can overlay optimization techniques to ask what the best or optimal offer we can make to any particular segment is based on goals such as maximizing customer profitability, loyalty, or any metric that we choose. We also can layer constraints (production capacities, distribution channel limitations, etc.) onto the optimization to make it far more practical and show an evolutionary path between today's reality and tomorrow's goal.
The following two case studies illustrate some of the many applications of the technique.
| Case Study: Discount Brokerage |
In the early '90s, a major discount brokerage had launched a low-price online trading subsidiary. All was fine during the dot.com boom, but as online trading matured (rapidly), the firm was grappling with a number of issues, including:
Approach
Online choice modeling involving current and prospective customers presented a wide variety of combinations of price, brand, and features. Survey respondents made choices between hypothetical types of trading accounts constructed dynamically by altering one or more of the variables. Optimization was then used to test the winning combinations against asset retention and profit potential.
Results
The insights helped the company decide to reintegrate the online subsidiary and establish single-tier prices at the lower price structure. Increases in account minimums for the new trading accounts were found to present no obstacle to consideration, yet ensured customer profitability. The brokerage became the first integrated online/offline trading service in the market. |
| Case Study: Electronics Retailer |
|
A major electronics retailer was rapidly losing share to a new competitor. Top management entertained many theories about the loss, each implying very different "fix" strategies. As executives couldn't just throw money (or focus) at all theories simultaneously, they sought to quickly test the following potential problem areas:
Approach
Choice modeling was used to draw links between store visit and purchase rates, customer experience satisfaction levels, and new concepts for in-store customer experiences. Various options of price competitiveness, merchandise mix, store layout, and customer service were tested to identify the upside potential of "fixing" each of the problem areas. The survey asked customers which efforts would most improve their shopping experience. By varying descriptions across many survey respondents, we could determine which functions would be most appealing and whether or not the whole idea would measurably improve customer experience.
Results
The choice modeling revealed that service interaction with the sales associates in the store was the single biggest opportunity for fast improvement. Merchandise mix and pricing were fine for now. Store layout and design could be improved, but they were not the primary obstacle at this point. As a result, a set of unassuming customer greeting and qualifying questions were developed, and the store personnel were trained to better diagnose the customers' preferences for more active or passive support and respond accordingly. In-store tests refined the most promising new service choreographies.
|
