Precima’s insight reached executives through analysts who decoded hard software for them. I proposed putting the intelligence in the product instead: one integrated suite first, then an AI dashboard that tells each person what matters and what to do. The record starts in 2017.
The product served one persona, the analyst. Everyone else got answers through people, and that friction was built into the business model.
Unify the tools into one suite, then build a single intelligent view that reads across the silos and serves each role directly.
The SaaS suite shipped and was core to the 2020 acquisition. The centralized dashboard went live on desktop, and My Precima became working software that answered in natural language.
Precima’s retail analytics were powerful and hard to use. Insight reached decision-makers through Precima analysts embedded with each client, who worked the tools and delivered the answers. It kept clients close. It also capped the value at the number of analysts, and it meant the product only ever served one person: the analyst.
That pattern is common in enterprise analytics. Complexity becomes the moat, and people become the middleware. My view was that the next wave of technology would break it, and that the company should be the one to move first.
The question I put to leadership: what if the product delivered the insight, and our analysts did the work only people can do?
The vision was not a late pivot. It is dated, in my own decks, from the first months of the job.
“My approach to making a great product,” and a price-optimization design challenge, before I was hired.
The first look at the product: Users need to see both for context… surface what the user has selected.
A tenfold larger audience beyond the analyst. More predictive, less reactive.
Less interface, more interaction… an assistant in their text window.
Pricing, promotion and assortment as modules in one environment, with contextual relationships across applications.
An innovation roadmap to 2022, built around one user request: Tell me what the problem is and how I can solve it.
Napkin sketches with the company’s president became a vision, user research, an architecture and a working software proof of concept, taken to key European clients.
Nielsen, now NielsenIQ, acquired Precima. I joined through the acquisition.
I led user experience, and used that seat to argue for a different product and a different business model.
The concept began as paper-napkin sketches with Precima’s president, Brian Ross. I turned it into the innovation roadmap and the My Precima vision, plan and concept design.
From gatekeeper to trusted advisor: a wider audience for the data would create higher-value work for client services, not remove it.
Some executives were all in, some on the fence, some against. I made the case room by room and built it with a small team, backed by data, sales and engineering leads who kept every phase feasible. It’s where I honed my boardroom skills.
Task-frequency research across retail and CPG roles, with client services as power users and as the voice of the occasional user.
Trained on Watson and built a contextual chatbot with a Python REST layer myself: an assistant that knew which page the user was on.
Led the future-state sessions with key European clients, Jumbo and Morrisons. Their buy-in unlocked a skunkworks build running in parallel to the main desktop replatform.
Each one moved intelligence out of people’s heads and into the product.
One suite with shared context across pricing, promotion and assortment came first. It shipped as the SaaS platform.
Why: an intelligent layer can’t read across tools that don’t share context. It was also the tactic I used internally to fund the bigger picture: integration gave clients a reason to buy the whole suite, and that revenue could pay for the centralized AI.
The target user became the decision-maker who never logged in: executives, category managers, merchants and sales teams.
Why: an analyst-only product capped adoption. The value was in the much larger audience the business had been serving through people.
A single view that reads across strategic levers, shows each role what matters now, and opens the evidence in one click.
Why: pulling a lever in pricing moves promotion and assortment results, but siloed tools were blind to each other. The decision lives across them.
A phased path from descriptive analytics, to a conversational interface, to prescriptive and automated recommendations.
Why: “Tell me what the problem is and how I can solve it” was the real request. Each phase delivered value with the technology available then, and was built to grow as compute and data costs fell.
From the 2018 concept to working mobile flows. Images are prototypes with placeholder data. Select any image to expand it.
Problem, rationale and solution in plain language, with a one-click path into the evidence in each tool.
A narrative view across the suite: what changed, why, and what to do next.
The concept from the December 2018 roadmap.Help non-analysts bypass siloed legacy workflow, in a format that fits their week.
“Problem, reason, solution.”Each answer earns the right to the next level of detail, so people can act fast or dig in. The same sequence is now a common pattern in AI analytics.
The recommendation in plain language, for this role, right now.
The context behind it: what changed, and which levers moved.
One click into deeper analysis in the tool that holds the evidence.
Phased so every step shipped value with the technology of its time.
A launch page for the suite, then a role-based executive dashboard of key KPIs, carried to mobile.
From KPIs to actionable insight, a conversational interface on mobile, and price and promotion scenarios.
One screen for everything: drill-down without leaving it, and an interface the AI generates for each person and need, with real-time predictive analytics underneath. No more pages, no unnecessary workflows.
The mobile app turned the conversational idea into flows people could test.
“Show me the sales of baby natural foods.” “Show me its baseline sales.” “Pin this KPI to my dashboard.”
The assistant works out what “its” refers to from the conversation.Tap a KPI to move from category totals to the items behind them.
Depth stays one tap away.Who needs what, how often, and how it connects to the platform underneath.
Executives, category managers, analysts, vendor advisors and sales, mapped by how often each task happens.
The occasional user was the opportunity.Mobile and suite interfaces over new cloud services for language and conversation, connected to the existing analytics engine.
New services on top of what already worked.The same KPIs and language from the suite to the phone.
The market arrived at the same ideas years later. These are parallels, not influence: I’m not claiming anyone drew on this work.
Put the analysts’ work into the product, and stop selling access to experts.
Foundation Capital names the shift service-as-software: AI sells the work, not the tool.
My Precima: personal KPIs, an alert when something significant happens, insight in plain language, sharing, mobile first.
Tableau Pulse: personalized metrics with AI summaries, delivered in Slack, email and mobile. Previewed May 2023, generally available February 2024.
A conversational assistant for business users that knew which page they were on.
Databricks AI/BI Genie: conversational analytics for business users.
An interface generated for each person and need, with no fixed pages.
Google’s Generative UI: an interface built for each prompt, in Gemini and Search.
Natural-language analytics predates this work. What ran ahead was the combination, and the argument for changing the business model around it.
I made the case for what investors now call service-as-software in 2017, seven years before it had a name.
The suite and the centralized dashboard shipped. My Precima was built, working and never sold: it ran ahead of the business model it landed in, and ahead of the cost curve that would make it easy. The direction outlived both.
“Tell me what the problem is and how I can solve it.”