Legacy spreadsheets and data systems continue to frustrate the pursuit of advanced analytics for many organizations, despite it becoming clearer than ever that agentic AI is the future of retail analytics.
Agents that not only answer the questions being asked of them, but that also guide users through defined analytical workflows. Agents built on trusted data, with defined business rules and repeatable frameworks, can now recommend the next-best-action, finally moving teams away from descriptive analytics to prescriptive and predictive models.
Ultimately, retail analytics has always been about answering a simple question: what is happening in the market, and what should we do next?
In all our conversations with clients, it’s been consistently clear that it’s always been a challenge to reconcile multiple data sources, ensuring accuracy and confidence in the answers analytics offer. This challenge has only become more acute with the advent of AI and the ever-expanding quantity of data available from various sources.
For many CPG insights teams, the journey began with spreadsheets. Retail sales, market share and distribution data had to be manually cleaned and painstakingly combined before analysis could begin. These tools met a need and made data analysis accessible, but they also created familiar problems: inconsistent definitions, duplicated effort, limited traceability and too much time spent preparing data rather than interpreting it.
It also became clear, as those spreadsheets proliferated, that scalability was nigh on impossible, and as more teams and users utilized the data, the possibility of file corruption and inconsistency added more problems than the tool sought to solve.
Add to this the fact that the data was static and structured in a way that meant it addressed a single question or served a single use case. There was no real possibility of diving deeper into the data and very little context to explore outside the spreadsheet’s pre-set parameters. The ability, therefore, to answer follow-up questions was left to the creation of yet another spreadsheet or, worse case, taken “offline” and man-handled into some sort of answer.
The first major milestone was the adoption of business intelligence platforms. Dashboards made recurring reporting faster, enabled certain data cleansing tasks to be built into the report generation itself (as long as the data inputs remained the same) and supported a certain amount of “drill-down” if follow-up queries were asked.
Data warehouses then brought larger datasets together, supporting more consistent analysis across brands, categories and markets, managed by an in-house data team who could maintain data quality and define the “single source of truth”. But as with business intelligence platforms, if data inputs changed, rules that had been put in place to maintain data integrity could break, and it required constant vigilance to ensure rules reflect the revised inputs.
The growth of e-commerce, digital shelf and retail media added another turning point, as CPG companies needed to connect increasingly granular data across physical and digital channels.
Today, retail analytics is entering a new phase. Cloud platforms make it possible to store and process data at far greater scale. APIs enable information to move more quickly between retailers, data providers and internal systems. AI creates opportunities to identify patterns, generate forecasts and support faster decision-making.
However, technology adoption does not automatically create AI readiness. In our recent LinkedIn poll, 63% of respondents said they have some AI foundations in place. Yet 13% remain reliant on manual spreadsheets, while 19% are restricted by legacy systems. Only 6% feel confident in their organization’s AI readiness.
It is likely that the number of companies still reliant in some way, shape, or form on legacy spreadsheets and legacy systems, especially in smaller or more local teams where centralized systems may not accommodate the local nuances or context. These highly specified and tailored data processes are also the hardest to replicate at scale, adding a further dimension to the challenge facing global organizations.
However, given that almost 2/3 of respondents claimed to have some AI foundations in place, this does show an industry moving forward, but not yet operating from a consistently strong foundation. Limited by their legacy tooling, preventing them from keeping pace with the opportunities AI presents.
For CMI leaders, the next evolution will not simply be more dashboards or more models. It will be analytics that is easier to access, more connected to business context and increasingly embedded in everyday decisions.
Generative AI allows users to ask questions of their data in natural language, while predictive and prescriptive analytics help teams anticipate change and recommend action. This shifts CMI teams from retrospective reporting towards continuous, real-time intelligence: understanding not only what happened, but why it happened and what comes next.
Agents can guide users through defined analytical workflows, supporting use cases such as performance diagnosis, opportunity sizing, root-cause analysis, and data quality audits, not as one-off snapshots but continuously, in as close to real-time as the data availability allows.
But these capabilities depend on reliable inputs. AI cannot compensate for inconsistent product hierarchies, mismatched time periods, changing retailer definitions, or fragmented market data. Without context, even sophisticated models can produce outputs that are difficult to explain, compare, or trust.
This is where Redslim comes in.
We have always sought to provide our clients with stable, harmonized data inputs that can support their BI ambitions. Now, in the AI era, we are supporting the agentic evolution by creating data products that align to our READY standard: Reliable, Explainable, Adaptable, Discoverable, Yours.
Infinitely composable, every data product Redslim creates can be used alongside its siblings, ensuring that the models created are speaking the same language and abiding by the same definitions.
The machine-readable context layer that sits within each of these data products provides delivery, processing, lineage, and safe-use instructions, with our playbooks helping to embed guardrails and analytical boundaries directly into the workflows. Users can explore the opportunities enterprise and AI-READY data can offer, while ensuring that business rules, definitions, and calculations remain consistent across both Global and Local teams.
Let us help your organization harmonize complex, fast-changing data and add the context required for reliable analytics and AI. Get in touch to explore how we can help.
