Learn more about our Innovator B2B Growth & Marketing Services
Databanks deliver streamlined access to high-value data, integrating Coresight’s proprietary research with trusted external sources to offer comprehensive coverage of the consumer, retail, technology and real estate landscape. By centralizing diverse datasets in a single, searchable environment, the platform reduces research time through with ready-to-go time-series data sets.
Retailer Dashboards introduce a powerful new way to access essential data and insights on key companies, delivering everything users need in one intuitive, regularly updated tool. Each dashboard combines a seamless interface with robust data to deliver actionable insights that empower informed decision-making.
The Store Intelligence Platform (SIP) is your always-on tracker of U.S. retail activity. It monitors store counts, inferred openings and closures, and precise store locations—delivering a real-time view of the retail landscape. Currently in Beta, the platform is continuously evolving with new features and expanded coverage.
More information on Store Intelligence Platform data sourcing and methodology
Weekly ML-generated sales projections across 85 consumer packaged goods models, powered by real-world sales data and Coresight’s proprietary consumer survey signals.
AT A GLANCE
CPG Sales Projections is a Coresight Research data product that delivers timely, accurate forecasts of US in-store sales across the consumer packaged goods universe. Designed to help retailers, manufacturers, and financial institutions stay ahead of fast-moving market dynamics, it provides a 12-week rolling forward view, updated every week.
CPG Sales Projections integrates Coresight’s exclusive weekly consumer survey data as a leading demand indicator, capturing shifts in purchasing behaviour before they appear in the sales numbers. Proprietary machine learning models then combine both inputs to produce forecasts that are meaningfully more accurate than traditional approaches.
Categories
Weeks Ahead
Updates per Year
Performance Views
Forecasts span 73 CPG sub-categories organised into 11 categories, plus a total CPG model, 85 models in all, covering everything from beauty and beverages to pet care and tobacco.
Absolute sales forecasts are generated for each of the 12 coming weeks and refreshed every week, so clients always have the most current forward view available.
Dashboards display both year-over-year (YoY) and week-over-week (WoW) changes, enabling quick identification of growth areas and category risks.
Data represent in-store sales across food retailers, drugstores, mass merchandisers, warehouse clubs (excluding Costco), dollar stores, military retailers and convenience retailers.
CPG Sales Projections combines two data sources, each updated weekly:
Proprietary machine learning models trained on both inputs generate weekly forecasts across all 85 categories. Four algorithms are evaluated on every category simultaneously, each designed to capture a different pattern type — long-range trends, short-term volatility, feature interactions, and seasonal cycles. The best-performing model is selected per category based on which one most closely predicted actual sales outcomes in recent weeks, ensuring optimal accuracy regardless of how a given category behaves.
Our ML modelling has validated that survey data serves as a leading indicator of CPG sales, and that forecasts produced with survey input are meaningfully more accurate than those produced without it.
We rigorously test the value that our consumer survey data adds to forecast accuracy. The results are consistent and significant across every category we track.
Survey-augmented models consistently deliver 95%+ average forecast accuracy across tracked categories. The survey signal is a validated leading indicator: it picks up demand shifts before they appear in the sales data, and the accuracy gains are consistent across all categories tested.
Accuracy Testing ResultsSurvey-augmented forecasts consistently outperform models run without survey input. The charts below illustrate that performance across five categories, showing how the survey signal improves both accuracy and stability week over week.
Drive smarter merchandising, promotional planning and shelf strategy with accurate forward-looking category intelligence updated every week.
Optimize production planning, inventory management and distribution by understanding where category demand is heading, not just where it has been.
Build a more granular, data-driven view of consumer spending trends to sharpen their investment thesis and anticipate sector movements ahead of the market.
Total CPG · 11 categories · 73 sub-categories
85 models across 11 categories
85 models: 73 sub-categories across 11 categories, plus category totals and Total CPG.
Sample output from the CPG Sales Projections dashboard showing YoY % change, by week
Illustrative output based on CPG Sales Projections methodology. Actual dashboard available to subscribers.
Forecasts are refreshed every week. Each update incorporates the latest available sales data and the most recent consumer survey results, so the 12-week forward view always reflects current conditions rather than lagging behind market moves.
The underlying sales data covers: grocery retailers (including supermarkets), drugstores, mass merchandisers (including Walmart and Target), warehouse clubs (excluding Costco), dollar stores, military retailers and convenience retailers.
Our weekly US consumer surveys record the percentage of respondents who reported making in-store purchases across a range of categories in the prior two weeks. Because consumer intent tends to shift before actual spending does, these survey responses serve as a leading indicator, giving the model an early signal of where demand is moving. ML modelling has validated that including this survey input produces forecasts that are meaningfully more accurate than those generated from sales data alone.
The 85 models we run cover very different data characteristics. Some have slow-building multi-week trends; others are dominated by sharp seasonal spikes; others show complex non-linear relationships between variables. A single model architecture cannot capture all of these patterns equally well. By running four algorithms simultaneously, each specialising in a different pattern type, and selecting the best-performing one per model, we ensure that forecast accuracy is maximised regardless of how a given category behaves.
The CPG Sales Projections dashboard presents forecasts as percentage changes, both year-over-year (YoY) and week-over-week (WoW), making it straightforward to identify which categories are accelerating, decelerating or tracking in line with seasonal norms. Users can view projections at the total CPG level, by category, or by individual sub-category.
Please contact us to discuss access options. Our team will be happy to explain how CPG Sales Projections fits within our broader suite of data products and memberships.
Coresight Research helps clients accelerate innovation and growth. Submit your information to schedule a call with a member of our team to learn more about how our unique service offering can benefit your business.
