In this article, I discuss the internal processes of analytic consulting firms - companies that use data-driven solutions to help their clients. Such firms are not primarily focused on statistical modelbuilding or even data analysis. Rather, like all management consultancies, their concerns are focused on understanding business challenges and delivering actionable solutions. In addressing these concerns, there are crucial data gathering and analysis phases which involve thorough data exploration, statistical modeling and usually a fair amount of technical sophistication. But these processes are just the means to an end. This article focuses on the methodologies that Inductis applies to ensure the quality of these intermediate processes.

Consistent delivery of effective client projects depends upon three critical components:

People
The key element of any consultancy is its people. Business consulting requires a passion for solving business problems (and commensurate expertise and insight) along with a strong commitment to teamwork. But effective analytic consulting requires additional specialized skills, including:

• statistics
• econometrics
• operations analysis
• information technology
• data mining

Inductis relies upon a rigorous recruitment process and maintains a strong commitment to ongoing training in order to meet the needs for cutting edge skills in all these areas.

Tools
Specialists cannot deliver full value without a diverse toolkit organized for efficient access. Key components include:

• A wide range of analytic software (SAS, CART, SPSS, Excel etc.)
• A searchable library of macros for the efficient execution of common complex analyses
• Proprietary analytic tools that extend the capabilities of available commercial software
• A high speed data network with substantial redundant storage capacity and processors
• Rapid access to common overlay datasets
• Extensive documentation for these datasets
• Searchable archived documentation of technical breakthroughs from previous projects
• Sharing of skills and analytic discoveries through internal reports and training programs
• Professional development through participation in conferences and seminars, white paper publication, conference presentations and articles

Process A consistent well-planned businessprocess is as important for a consultancy as it is for its clients. Critical as they are, people and tools are insufficient without the right business process. Inductis has developed the MicroAnalytix© Process (Figure 1) to ensure consistent delivery on its promise to “enable clients to make the best possible operating decisions by developing effective analytic solutions.”

The MicroAnalytix© Process

Inductis uses the MicroAnalytix© Process as a general framework for its commitment to enable companies to make the best possible operating decisions by developing effective analytic solutions and operating them efficiently. MicroAnalytix© provides the general framework for project planning, while specific models and analytical techniques are selected according to the unique requirements of the project. There is no one-size fits all.

QUALITY ASSURANCE
Project excellence depends upon the participation of the right people with the right training and consistent application of the MicroAnalytix© process. Inductis internal QA process includes four Quality Assurance Checkpoints (Figure 1) that ensure:

• the systematic application of MicroAnalytix©
• the use of analytic models grounded in sound statistical, econometric and datamining practices
• the delivery of reliable analytic solutions for business clients
• freedom from errors in model specification, coding, transcription and variable interpretation
• efficient capture of analytic learnings to enhance future project effectiveness

The QA process is a systematic overlay designed to ensure the application of consistently high methodological standards throughout project design, development and delivery. A core requirement is that every project team is followed by a senior Inductis methodology leader who serves as methodology reviewer and advises the team as needed throughout the project. In particular, the reviewer is responsible for formal sign-off at the four QA checkpoints.

THE MicroAnalytix© CONSULTING PROCESS

DESIGN – Understanding the Strategic Imperative
Consider the example of a credit card issuer who has a concern about loss rate (write-off as a percent of total revenue). This could result in an analytic consulting project focused on building a model to predict losses based on application and bureau data, with the idea that this model will be used in decisioning on new applications. Once the project has been focused on this relatively-narrow task, the issues during the project would include data availability at time of decisioning, ensuring availability of the correct bureau file for the historic sample, missing data imputation, workarounds for application data changes over time, identification of a relevant validation sample etc. Once the best possible write-off prediction model had been built, the joint consulting and client team would then turn to the issue of implementation and the question would arise: “How can we use this model to improve the loss rate?” It would become readily apparent that there were critical issues that the modeling project didn’t address: How to distinguish between high and low revenue (spend and revolve) clients at given levels of the new risk score; How pricing might be used to mitigate risk; How to decision around previously rejected applicants who would be approved under the new score; How to increase portfolio spend while possibly decreasing the number of accounts etc. Clearly, if the strategic concern were loss rate, then a project that focuses on loss incidence would provide only part of the answer.

Experienced management consultants and client executives play the lead role in addressing the issues of project objectives and scope raised in the above example. Inductis methodology leaders (statisticians and economists) also assist in developing possible implementation strategies.

Essential questions in the Design phase

• What is the core business challenge?
• How can this challenge be addressed (in whole or in part) through data-driven processes?
• What data is available?
• What resources (time and budget) can the client allocate at this time?
• What subsidiary answers would provide worthwhile
  actionable secondary targets?
• What initial hypotheses do we have to guide subsequent analysis and modeling?