Statistical analysis of survey and customer data, automated systems that collect it, and engineering that delivers it into client data environments.

• Advanced Statistical Analysis
• Automated Survey Data Collection Systems
• Data Engineering and Machine Learning
• Research Services

Advanced Statistical Analysis

Conjoint and discrete choice, MaxDiff, segmentation, driver analysis, pricing models, and predictive modeling. Engagements cover experimental design, estimation, and delivery of results as reports, data files, or working tools such as market simulators and segmentation typing tools.

• Conjoint and discrete choice - experimental design, hierarchical Bayes and latent class estimation, and market simulation.
• MaxDiff - best-worst scaling for features, messages, and claims, with individual-level scores.
• Segmentation - k-means, hierarchical, and latent class methods, with segment profiles, sizing, and typing tools for classifying new respondents or customer records.
• Pricing models - conjoint, Van Westendorp, Gabor-Granger, and revenue modeling.
• Key driver analysis - regression and relative importance methods for satisfaction, loyalty, purchase intent, and other outcomes.
• Factor and path analysis - reduction of large attribute sets and structural models that trace how perceptions act on outcomes.
• TURF - reach optimization for product lines, flavor sets, and message portfolios.
• Machine learning - classification and predictive models built for interpretation as well as accuracy.
• Weighting - raking and post-stratification to population targets, with design effects reported.

Simulators and typing tools are delivered as spreadsheets, web tools, or code that runs in client systems. Descriptions of each method are on the full capabilities page.

Automated Survey Data Collection Systems

Continuous trackers, CX and Voice of the Customer programs, employee listening, and recurring institutional studies, run on the Survey Launch Automated Data Collection Framework. Once deployed, the pipeline runs unattended. Survey programming, content and asset preparation, approval gates, and quality review are handled through a portal.

• Automated collection - surveys launched by triggers, clicks, or schedules.
• Automated processing - data blended, normalized, and harmonized with external sources.
• Automated analysis - NPS, classification, complex models, and structural equation models.
• Flexible delivery - portal, push, ERP integration, data visualization tools, API, and Model Context Protocol (MCP).
• Consistent execution - framework rulesets run identically on every wave, with a full audit trail and logging.
• Cost structure - cost tracks deployed modules, not seats, users, or responses. Additional waves, markets, and projects add compute rather than headcount.
• Vendor-managed infrastructure - infrastructure as code (IaC) cloud deployment. No software to license, administer, or staff.

A media survey platform built on the framework has operated continuously since 2009. As of September 2026 it has managed 41,000 projects in 13 countries, with peaks above 400 projects per month. A weekly tracking study has run on the framework since 2006 and as of September 2026 has been fielded 1,063 times. Neither has had a programming error or a missed launch. Both are described in the case studies.

Data Engineering and Machine Learning

Survey data carries structure that general-purpose pipelines do not model well. Question and value labels, multi-response items, skip logic, weights, and instruments that change between waves all have to be preserved in the warehouse schema. Engagements range from a single integration to ongoing operation of pipelines and models in client environments.

• Pipeline engineering - ingestion, transformation, and scheduled delivery of survey data to Snowflake, BigQuery, Databricks, Redshift, and other platforms.
• Survey data modeling - warehouse schemas that keep labels, metadata, weights, and wave history usable by analysts and BI tools.
• Data integration - survey responses combined with transactional, behavioral, and BI data and consumer metrics, with matching, keying, and documentation.
• Model deployment - predictive models, classifiers, and segmentation typing tools deployed into client environments, with monitoring, scheduled refreshes, and re-estimation as the data changes.
• APIs and web services - real-time data exchange between survey systems and client platforms.
• Custom applications - data collection platforms, secure portals, and applications for work that standard survey software does not handle.

The automated collection systems above are built from the same pipeline and delivery components.

Research Services

Design and execution of complete survey projects, or any part of one. The same research services support the analytics and data collection work above.

• Research design - method selection, measurement, and the analysis plan for a stated question.
• Sampling design - stratified, multistage, and multilingual designs, including approaches for hard-to-reach populations.
• Questionnaire design - instruments that measure what the analysis plan requires, with attention to question order, scales, and respondent burden.
• Survey programming - complex instruments with conditional logic, randomization, quotas, and multilingual versions, fully tested before launch.
• Fielding and field management - launch, monitoring, quota management, and response tracking across online, telephone, in-person, and mixed-mode studies.
• Sample management - sample from real, known sources, including client customer lists, recruited panels, and partner sources, coordinated across providers with deduplication and quality screening. Programmatic marketplace sample is not used.
• Tabulation - cleaning, weighting, and banner tables with statistical testing.
• Analysis and reporting - plain-language reports, presentations, and data files.