Retail and consumer goods teams make decisions across a connected but complex operating cycle. A Monday morning review may begin with sales, inventory, transactions, traffic, and delivery KPIs, then move into category performance, demand forecasts, campaign planning, and store execution. The signals may come from point-of-sale, loyalty, supply chain, media spend, and store operations, but the decisions still need to line up around the same products, categories, and locations.
Those decisions rarely belong to one team.
- A category planner may need sales and inventory signals to adjust a forecast.
- A marketer may need those same signals to identify an audience and shape a campaign.
- A store manager then needs clear tasks to execute the promotion, while leadership needs to understand whether the activity improved performance.
When these decisions are spread across disconnected systems and teams, it becomes difficult to turn analysis into results. A unified application can connect these steps in one end-to-end retail process.
In a Databricks demo presented by Pavi Singh, Senior Solutions Architect for Retail and Consumer Goods at Databricks, stakeholders can use a single application to review performance, plan demand, activate audiences, coordinate store execution, and measure campaign outcomes.
Three foundations for a connected retail workflow
The application is built on three foundations: data unification, governance, and intelligence.
Data unification creates a single source of truth across signals such as point-of-sale, loyalty, supply chain, media spend, inventory, and store operations. Instead of asking each team to reconcile separate views, teams can work from a shared foundation.
Governance controls who can access different types of data and how that data is used. This matters when executives, marketers, planners, store managers, and store associates use the same application but need different views and responsibilities.
Intelligence supports recommendations and automated decisions. In the demo above, Genie enables natural language querying, while autonomous agents help orchestrate multiple steps across sales insights, demand planning, audience building, in-store operations, and measurement.
Together, these foundations connect the data, controls, and decision support required for an end-to-end retail workflow.
Start with a shared view of retail performance
The workflow begins with a performance view for roles such as a CMO, marketing director, or category planner. They can review sales, inventory, transaction count, traffic, and delivery KPIs, then examine monthly sales trends and performance by category.
From there, users can drill into categories and subcategories to identify top movers and underperformers. An executive brief can summarize the current position, while more detailed views help practitioners understand what is driving the numbers.
The same application also provides a map view for store-level analysis. Users can review on-hand stock, units sold, revenue generated, and the stores contributing to performance. This gives teams a path from an overall KPI to the specific stores, categories, and products that require attention.
Move from forecast review to planning decisions
Once a category has been identified, the next question is whether demand is tracking to plan. The demo shows forecast-versus-actual analysis across a rolling four-month window, with the ability to move from category-level trends into subcategories and individual SKUs.
Practitioners can then examine daily divergence rates, volatility, and gaps across a month or season. This adds context to a forecast variance by showing when performance changed and what may have influenced that change.
The workflow also supports planning actions. Users can review recommended reorder units for a specific stock-keeping unit (SKU), category, or store. With what-if scenarios, a merchandiser or planner can adjust a business lever, such as a discount, and observe the projected impact across replenishment, audience activation reach, units, revenue, and other KPIs shown in the application.
The goal is not simply to report what happened. It is to give practitioners a structured way to test potential decisions before acting on them.
Turn demand signals into audience activation
Demand and sales signals can also inform marketing decisions. In the demo, an audience wizard recommends potential campaigns based on the signals identified in the application. Examples include a food storage recovery campaign and other campaign ideas surfaced from the data.
The campaign workflow guides users through defining intent, clarifying KPIs, selecting recommended audience segments, specifying spend, adding creative details, and activating the campaign. Users can review the expected audience size, average ROI, activation window, and other campaign details before launch.
This connects planning and marketing in a practical way. A category or demand issue does not remain isolated in an analytics view; it can become an audience and campaign decision that is tied back to the underlying signals.
Bring the workflow into the store
After a campaign is activated, the work continues in the store. The demo shows a mobile-oriented view for a store manager, including restocking units that need to be delivered across SKUs, price-tag updates, task assignment, and campaign-related promotions.
Store associates receive a checklist of assigned work and can track completed tasks. The application can also surface recommendations for promotions or replenishment, helping associates understand the operational actions associated with the broader campaign.
This creates a shared workflow across roles. Planners and marketers can define the commercial action, store managers can assign the work, and associates can execute and track it at the store level.
Measure what happened after activation
The workflow closes with campaign performance measurement. Users can return to a category such as food storage recovery and review the metrics associated with the campaign, including recovery velocity, ROAS by tactic, featured-store insights, and other performance measures presented in the application.
This closed-loop view connects campaign activity back to sales and operational outcomes. It helps teams move beyond launching a campaign to understanding how it performed across categories, stores, and tactics.
From fragmented signals to connected decisions with Genie
Retail and consumer goods teams need more than another dashboard. They need a way to connect performance analysis, planning, activation, execution, and measurement without losing the context or controls required at each step.
The demo shows how multi-agent Genie can coordinate this work across personas and application areas. A user can ask it to plan a food storage category recovery end-to-end. The resulting view brings together sales, margin, in-stock rate, critical issues, portfolio actions, inventory, and supply chain replenishment, and operational recommendations.
Behind the scenes, the application uses agents focused on different parts of the process, with a supervisor coordinating the overall workflow. The architecture view shows how data powers the individual Genie agents and how the supervisor connects them across sales insights, demand planning, audience building, in-store operations, and measurement.
For practitioners, the important point is the design: each step can remain connected to governed data and its operational context while agents help coordinate work across functions. Teams can start with a KPI, investigate a category or SKU, test a planning decision, activate an audience, coordinate store tasks, and measure the result in one connected process.
For teams evaluating how to connect retail workflows, the practical lesson is clear: start with a shared data foundation, apply governance across personas and data access, and design intelligence into the path from insight to action.
Looking to learn more about what Databricks can do for you? Check out the resources below:
Frequently asked questions
What is connected retail demand planning?
It links sales, inventory, supply chain, marketing, and store signals in one workflow, helping teams move from demand changes to forecasting, what-if planning, activation, execution, and measurement.
How can unified data improve retail decision-making?
It gives planners, marketers, store teams, and leaders a shared view of products, categories, and locations, with consistent KPIs and a clear path from business question to action.
What data sources does a unified retail application typically connect to?
It typically connects point-of-sale, loyalty, supply chain, media spend, inventory, and store operations data. A shared foundation helps teams move from KPI review to category analysis, planning, audience activation, and campaign measurement without reconciling separate views.
How does AI support retail store execution?
AI can help translate campaign and planning decisions into store-level actions, such as replenishment, price-tag updates, promotions, and assigned associate tasks.
How does Genie support retail teams working with complex data?
Genie lets retail teams query governed data in natural language and coordinate work across sales, planning, audience building, store operations, and measurement, allowing a practitioner to initiate an end-to-end category workflow from a single prompt.
How can retailers measure campaign performance?
By connecting campaign results to sales, recovery velocity, ROAS, and featured-store performance. This closed-loop view helps teams understand what worked across categories and stores, and improve future planning and activation decisions.
Instagram Engineering
Shopify Engineering