Case study

Ipsos Retail Performance

Personalised MI point-of-sale reporting

The scope for a significant project

During the Discovery Phase of this project, we were asked to consider a totally blue-sky approach, where the project would involve delivering a totally new client design/build and an entirely re-architected database and associated server solution to deliver more flexibility and introduce options for the future.

Such a project would involve significant testing, planning, resources (both development and Ipsos technical), rollout management and migration work.

A Gantt Chart of our proposed project plan with key milestones exposed is shown here.

Gantt Chart for the project

The problem

The emphasis was to overcome all of the limitations of the existing system, redesign the solution and the UX from scratch and, finally, wipe the slate clean in terms of existing server-side architecture, as necessary

The approach

To deliver all requirements in two low-risk phases that considers rollouts to 1000s of daily users... Phase 1 delivered a refactored client front-end with new development work undertaken to virtualise the reporting functions (including the data retrieval from NOVA/Cube/other sources) to deliver an updated solution in short order. Phase 2 would run in parallel and extend the architectural tasks started by replacing the existing database, while extending the underlying data architecture to include new, third-party data sources such as Zapier, Salesforce, Social and information from digital marketing platforms, customers' own SOP, etc

The outcomes

Delivered the opportunity to integrate additional Ipsos sourced and external custom data sources into client reporting functions; providing more value and choice to customers. Enabled new functionality around predictive and exception reporting in combination with the use of machine learning to deliver valuable new insights, benchmarks and recommendations to customers. The existing system’s limitations were overcome and Ipsos now has a framework to continuously generate additional revenue

Ipsos Retail Performance
Visual design

Visual design

Visual design was key to ensuring that the system securely provided exactly the right data at the right time to the right people. To this end we came up with some very innovative and powerful ideas, as you can see from the accompanying, annotated screen grab.

Particular attention was paid to presenting complex data sources in meaningful ways and allowing users to save their own specific combination of reports as a 'dashboard'. Where each report can be based on querying any data set over any period of time, thus a Regional Manager could review weekly sales Vs footfall for all of her retail stores, whereas a Store Manager could review a daily dashboard for each of his store's departments and also review heatmaps showing dwell time on each floor of his store.

The new solution was fully responsively designed, meaning that regardless of the size and type of device that the user is using, the solution automatically configures its layout accordingly. If the user has the solution open in a large browser window on a 4k display, this can show 4 full detailed reports side by side. Whereas, on a mobile phone in portrait mode, the charts are stacked on top of each other (example given further below).

Software and data design

Key to any 'design process' is to understand your data...

Data design is therefore the essential precursor to approaching any visual or software design processes.

We created layers of related information that are rapidly accessible, mappable, comparable, presentable and meaningful. We designed multi-dimensional database processing to validate and aggregate continuous feeds of data to generate daily, weekly, monthly, quarterly and yearly dimensions of secure customer data.

An important concept in the new solution was to consider that the existing pre-defined reports (that users are familiar with) simply now become ‘encapsulated’ reports that are pre-configured, such that all the information needed to present them is self-contained.

In the new system, a named report is an object containing:

  • Data source access method, a Phase 1 re-factored stored procedure or 3rd party source
  • Telerik report or table generator to use (Phase 1) or the provider of the report UI, i.e. a third-party virtualised visual plug-in
  • All filter settings associated with the report Name
Software and data design
Data presentation

Data presentation

The presentation architecture

The entire content of the dashboard is easily and intuitively configured by the user, and all dashboards (and embedded reports) can be named and saved. This allows the user to create as many dashboards as they wish. This is particularly useful and an order of magnitude easier to use than the old UI, which forgets all reports’ settings when the user logs out.

The user is able to add, remove, edit and drag and drop to re-order the reports shown on any dashboard page. Once the user has created their preferred dashboards (based on either new dashboards or pre-configured dashboard templates that could ship with the system), they will only need to load them to gather their information and insights. It is also very easy to duplicate a dashboard to make an alternative version. We would replace the logged-in home page with the user’s preferred dashboard, so they can access their most important information immediately upon logging in.

Finally, the user is also able to select up to three headline values to display at the top-right hand area of the screen, showing their preferred metric, over selected time period and compared to previous metrics from their selected time period. The metrics, percentage differences and positive and negative comparison indicators are automatically added to the headline.

Ipsos Retail Performance

Example report


This was a large and complex technical and UX-design project. However, the extensibility and flexibility provided allowed users to interrogate any data and save multiple reports as dashboards that can be duplicated, automatically executed/emailed and shared with colleagues.

Finally, it also allowed Ipsos to add new external data sources meaning that there were no limitations remaining within the solution whatsoever. That's precicely what we were commissioned to design and deliver...


Chris Rodbourne
S-Digital, MD