AI Forecasting
Platform

Foresient is a state-of-the-art platform that enables demand planners to forecast at scale.

Foresient forecasting platform uses a suite of AI-based sophisticated algorithms that's equipped to handle varied real business scenarios, for instance predicting future sales.

The platform builds forecasting models by looking at historical data with a combination of other aspects like promotion, holidays, and schemes, among others. The engine runs multiple built-in algorithms, automatically identifies the best fit model, and predicts the time series. These meaningful outcomes can eventually help businesses to optimize the plans for the future.

My role: Senior Interaction Designer.

Duration of the project: 4 months

Demand Planner is someone who is there to drive the demand and inventory levels of consumer goods(known as CPG). In other words, to maximize cash flows, and sales and services levels.

Forecast: A combination of Category + Geography + Product + A time cycle.

Forecast list: A collection of forecast for a given point in time.

KPI: The value that is forecasted for the time cycles.

Design that helps demand planners forecast the right volume levels for stores/manufacturers etc. across multiple locations.

Build a design system(i.e Component library) that can be scaled.

Help developers use design tokens to establish an accurate handover via Zeplin.

How do you deliver high forecasting accuracy across multiple categories today?

What is a high forecast accuracy in a highly volatile and seasonal product category for you?

How do you submit modeling requests as per the business needs currently?

What statuses of the modeling requests are crucial for you?

What are the different hierarchy levels you view your forcasts in?

Where would you click on this design, and what do you expect to happen?

Be Transparent:

Our UI should be clear as data can be messy. Our web app shouldn’t be. We minimize visual noise to maximize focus on actionalble data. At the end of the day it has to be understandable, but our users should be the focus.

Be Accurate:

Our web app helps users push their most validated data because that’s what we believe creates an accurate forcast. If our design has done its job, forecasts will feel like real time data

View current & previous forecasts:
‍
In here, the demand planner will be able to view current forecasts & previous(published) forecasts. If required, the forecast data can also be exported in excel so that any immediate changes can be made. The demand planner can also slice & dice the data to view specific information and publish the forecast if satisfied with the results.

Visualise forecasts:
‍
As a demand planner, it is very important for them to visualise trend lines of the forecasts to capture any sort of anamoly or deviation. With the “view trend” action, the user should be able to quickly view a snapshot of the forecast and toggle between different timelines(mainly 4/8 weeks).

View model:
‍
In forecasts, it is crucial to view what type of models are influencing the forecasts. Models are usually prepared by the data scientists & demand planners together, however, sometimes, demand planners are not in the loop. Hence, a dedicated area to view this information is useful to the user.