IBM PAIRS website homepage with a satellite view of a Caribbean region, showing islands and ocean under scattered clouds, with text promoting big data mapping.
A searchable high resolution data analytics platform.

IBM PAIRS Geoscope

Big geolocation data is just a few clicks away.

Roles: Research, Visual, Branding, Strategy

The brief

IBM Research approached us with this new and exciting mapping tool they had just created. The tool itself was barebones, but the significant impact it could have on how people use large mapping data to discover patterns was unimaginable.

PAIRS is a platform designed to take geospatial (location information) and temporal (time) data from sources such as maps, satellites, weather, drones, IoT devices, APIs (i.e. Twitter), and contextualize that data to develop pertinent analytics.

Our mission was broken up into two phases. In the first phase we had to help brainstorm other uses for the application outside the realm of typical mapping analytics tools. The second phase was to develop the web based user interface for PAIRS.

Four examples of PAIRS data and analytics maps showing drone and satellite data for the Netherlands, irrigation and evapotranspiration in Kenya, elevation in Napa Valley USA, and corn fields with rain and elevation in Iowa USA.
Snapshot of the dev-tool and the typical analytical use cases.

The first phase: exploring concepts

We met and collaborated with stakeholders and analysts to understand their expectations and come up with some design drivers to help us move forward with the different use case concepts.

Open notebook with hand-drawn project notes, sketches of dashboard concepts labeled A, B, C, diagrams of iteration cycles, and template concepts labeled A to J.
Overview of the project plan after our initial meeting.

We brainstormed quite a few ideas surrounding larger topics including:

  1. Education: There were two concepts: the first was the creation of a comprehensive interactive database on the history of global migration to understand our origins, history, and learn about our present, and the second was to create modules to be used for geography university classes.
  2. Journalism: A new way to visualize and contextualize news and events in their geographic landscape. This included a way for people who were in secluded parts of the world to tell their stories in hopes of reaching a larger audience.
  3. Social: There were two concepts. The first allowed you to view and share with your friends and family your mark, locally and globally, with your social media geodata. The second was a first response alert to people who were in your area in case of emergencies. The app would alert the authorities and other users regarding people in need of help.

We landed on Story of Me and News Explorer as the two concepts to further refine. We began with creating multiple personas, each with their own unique use cases and testing them against our concepts.

Concept A: Story of Me

For Story of Me we needed to accommodate two different uses cases. First, would be someone who was an experienced traveler and had many social media channels for us to plug in and generate a mapped journey. The second would be someone who has not necessarily traveled the world, stayed locally, and might not have social media, but wants to create a journey themselves.

User interface screen showing a map view of San Francisco, CA from 2011 to 2016 with an option to add an image and caption describing the location.
First iteration of Story of Me.
World map with blue circular markers indicating popular photo locations in 2016, alongside social media stats showing 126 photos posted and 23 places tagged.
Second iteration of Story of Me.
Interactive digital world map with glowing orange areas indicating most photographed places and blue and green circles showing user and friends' locations, with filters for years and various stats.
Final dashboard screen of Story of Me.
Travel summary dashboard showing a world map with travel routes and stats, highlighted post from Reykjavik, Iceland with photos, comments, and travel stats for 2016.
What you see when you hover over a traveled line.

The final concept was built to accommodate many social APIs, including those from travel sites such as Lonely Planet and TripAdvisor. The platform not only plotted your footprint around the world but that of your friends as well. It gave users a way to see relevant statics such as popular hashtags in local cities to discover trends, information regarding the distance traveled or countries visited, and also notified you on who else had visited that country. It contained many other features that users could keep coming back to.

Concept B: News Explorer

The initial idea started out as a disaster relief dashboard. It featured a scrubber to monitor information as it was progressing, local relief centers with their contact info, as well as a module to input custom data. In the second phase we shifted towards a geotagged news exploring site. You could move around the world and see the news you wanted to see with a vast selection of pre-filters.

IBM SmartMaps interface showing a globe with tracked locations, featuring news of Super Storm Sandy landfall 12 hours ago with options for 3D and VR views.
First iteration of the News Explorer.
IBM SmartMaps interface displaying a map with colored markers and circles tracking drought and sandstorms in Africa from Jan 19 to Jan 26, with stats on casualties, damages, and headlines from sources like CNN and NBC.
Second iteration of the News Explorer.
Dark-themed SmartMaps interface by IBM showing a timeline of armed conflict events from January 27-31, 2017, with a world map highlighting terrorist attacks and battles, and a popup featuring a man with the label 'Shots Fires' and links to Natural Disasters and Drought.
Final dashboard screen for the News Explorer.

The second phase: designing how you map data

For this phase we needed to provide a search-friendly way to create visualizations for data analysts. To meet this goal, we conducted several interviews with data analysts and developers to understand their process. Once we understood what steps and parameters needed to happen in order for the user to get a visualization, we began to wireframe ways we could make the process as seamless as possible.

IBM PAIRS interface showing data queries and dataset details next to a map highlighting Spain with a dotted border.
Browsing data sets wireframes.
IBM PAIRS interface showing query setup with spatial focus on Spain and Portugal on a map, data sets for Dayment Historical Weather, and options for setting queries on temperature data.
Setting the query wireframe.

After testing the wireframes, we noticed a few problems:

  1. The panel system was a good start but it needed to be a bit more flexible. Rather than trying to show everything, it could be refined to take you step by step though the process. Some inputs also could be revealed and then hidden once the information was filled in, as it was unlikely developers would update that specific parameter.
  2. Language was getting confusing. Everything you do is a query by definition, but within a query were other queries to define the top query. We made some changes by renaming “queries” to “projects” as it meant the same thing, and its own queries to "subquery."
  3. We learned that queries, now known as projects, had parameters that were not dynamic, meaning that they could not change compared to the subqueries within that project. So there needed to be a system that distinguished those elements.

After our learnings, we moved on to creating the final designs.

IBM PAIRS dashboard with a world map prompting to start first project and three example projects showing heat maps of South America, Europe, and a city.
Dashboard empty state.
Dashboard interface showing project thumbnails including a heat map of Spain, a world map for climate change, and three example projects with heat maps over South America, Europe, and a city.
Dashboard with projects.

You can start your first query by clicking the top "+ New Query Project" button or by clicking on "Start First Project." We provided example projects to inspire and help users familiarize themselves with the tool and how the outputs were created.

IBM PAIRS dashboard interface showing a world map with geographic regions and a sidebar for project parameters and query options including location search and selection tools.
Defining the area in which your data will create a visualization.
IBM PAIRS dashboard showing a project with a map of Spain selected and a date range filter from January 1, 2000 to January 1, 2017.
Defining the range of time for that visualization playback.

After creating a new project you have to set the project parameters which will remain constant throughout the query. When defining the area, you have the option to hover over countries or cities by zooming in, use the rectangular marquee, draw, or upload your own coordinates. After selecting the where, then comes the when. The time is so you can view a playback of the data as it changes over that predefined time.

IBM PAIRS dashboard showing Project 1 with Spain selected and date range from 1990 to 2000, highlighting Daymet 30 years historical weather dataset and data layer with a list of names on the right, alongside a map of Spain.
Selection of the Data Layer from a Data Set
IBM PAIRS interface showing Project 1 with Spain selected, date range from 1990 to 2000, a subquery for 30 years historical weather by Rose Casey, aggregation set to median and value filter greater than 2000, with a map highlighting Spain.
Setting the aggregation and filters for the subquery

The next step is to select a data layer from either predefined Data Sets or your own uploaded data source. To finish off the subquery, you then set the aggregation and filters.

Heatmap overlay of Spain and Portugal showing data intensity with red, orange, yellow, and blue colors on an IBM PAIRS dashboard.
Visualization based on the subqueries.
Dashboard interface showing Queries API access control, data type selection, query health monitor graph of API calls and PAIRS queries, and query refresh settings.
Viewing API calls.

Finally, once you have defined another subquery and run the results you get a visualization, which you can then take the data and import it into a powerful JS visualization tool such as MapBox or OpenLayers. You can also access the API and view its API calls and adjust accordingly.

Reception

PAIRS went public for enterprises to use for free back in March 2018. The reception since its launch has been positive.

When referring to the UI we created, TechCrunch said “…a web-based interface that makes it easy to select different layers, manipulate them and combine them to generate new queries.” Link to TechCrunch article.

It was also featured in Forbes, ZDNet, and other big tech magazines.

The concepts are currently featured in the tool as example projects. But most importantly they helped secure funding from the IBM executive team in order to continue work on IBM PAIRS.