i2 Analyze & Enterprise Insight Analysis

What is the i2 Portfolio?

The i2 portfolio (part of Safer Planet) produces visual intelligence and investigative analysis software for military, law enforcement and commercial agencies.

You know the old cop shows, where they pin people, places and crimes to a board and connect them with string? Imagine that on the web, connected to the criminal records database, the DMV and more, backed by specialised and highly advanced analytics to aid intelligence production and operational insight. Thats i2.
 

The Task

As a large legacy portfolio, our task is not quite the same as that of a design team in IBM, the i2 design team works split across the maintenance and updating of multipleproducts & solutions. My particular focus area has been in the creation and launch of our first hallmark project; the i2 Analyze Web Client, providing a thin web based intelligence analysis offering, drawing on and accounting for the needs of each of the existing offerings.


The Challenges

  • Acquistion: i2 doesn’t work like IBM, and has an instilled culture where design has traditionally been a tool to create particular features, rather than impact the product overall; “If your CEO says to do something, even if its wrong, it’s right” 
  • Complex Product. After over 20 years of steady functionality growth, and a large product portfolio, i2’s products have become exceptionally powerful, but built up incrementally. Leaving a complex and ‘doctorate’ level offering that is losing ground from its new UX centred competition.
  • Classified Users
  • Quarterly Release Cycles & Strict Feature Roadmap
    Not only is this a large and potentially bloated product set, the mission from the top is that the web client shall be delivered in quarterly cycles to act as a boost to sales of on premise services. The typical working method with design for i2 has traditionally been the common tale of quick fixes and releasing products based on their features alone. When we started this project, there was a clear ‘visual roadmap’ of which features should be released when.
  • Problem with this? Design started almost a full quarter behind.

 


Our Process

  • Workshops
    The first move by the design team to change the embedded culture in i2 to reflect our new hallmark status. Led by a lead, myself and one other designer we engaged a large portion of the i2 community and were able to reach 3 user centred hills and most importantly, defeat the feature roadmap approach in favour of a UX vision.

  • 10+ Users, 23 Interviewees
    With 20 years and a senior PLM previously a principle analyst, there is vast ingrained ‘legacy’ knowledge. We’ve never been on the web before, nor had new users to interview in a long time. In the 3 months post deployment we have interviewed users from 10 organisations in more than 20 sessions to bring research to the heart of our work.
  • Adopting Agile Practices
    As a newly dual located team, one of the first things I personally pushed through was a daily morning scrum of our design team. This bolstered the new collaboration events being employed by the team in the weekly and monthly interlocks to share the recent research and design work with the wider team (ask me for a deck!).
  • 1001 Prototypes
    Since deployment, I have personally created far more prototypes using a variety of methods (paper, keynote, axure) than I care to mention. When working at an abstracted level on exceptionally complicated topics, voice and static wires alone often fall short even for communicating designs. So I got scrappy and prototyped everything. 

Imogen

Imogen
All Source Analyst
-

“I need to use all of the tools at my disposal to deliver investigative insight to my team”

Imogen acts as a versatile ‘jack of all trades’ analyst, utilising all tools and data at her disposal across analysis types, focussing mostly on HUMINT (human intelligence), but dabbles across all other typical analysis techniques.

Typically receives RFI’s and outputs reports.

Moorad

Moorad
SIGINT Analyst
-

“I need to leverage huge amounts of signal data such as CDR from data and ad hoc sources”

Also part of the new breed of analyst, Moorad works in large scale enterprise agencies in the military or federal space. He specialises in leveraging giant data sets of signal data (call, cell tower pings, bank transfers etc) to uncover pattern of life and investigative leads for his team.

Receives RFI’s and outputs reports & support.

Sam.jpg

Sam
Principle Analyst (Manager)
-

“I need the ability to watch over and manage my teams of analysts and my missions”

Sam represents what is left from the days of i2’s proliferation and ‘king of the hill’ status. Having undertaken training in i2 software he is both an expert user and a massive proponent of i2 in his organisation.

  • Sam receives direction from upper management.
  • He is the one who starts and manages missions.
  • He tasks his team of analysts and reports its progress upwards and outwards.
Ryan.jpg

Ryan
Investigator (On the streets)
-

“I need a place to store all of this information aside from my notebook and somewhere to bring it back so I can better collaborate with the analysts”

In the federal and law enforcement spaces, the analysts aren’t alone, special investigators and detectives are the ones who create the data for each analyst to use. Not technically minded these users would find extreme benefit from a ‘thinner thin client’ of basic functionalities.

  • Ryan inputs his data into the system through custom forms.

Effort one: Day to Day UX

Screens.jpg

Quarterly Support

So running with a quarterly release we were in danger of designing feature by feature without knowing the overall context in the user’s experience or our end experience goals. To meet this, a lead and myself formulated a 3 ‘day’ experience scenario for our All source analyst from very basic usage using the most basic analysis techniques (day 1) and bringing in higher level usage abilities such as M.O. comparisons and basic geo searching (day 2) and call data records (day 3). 

Establishing & Stablising the End to End Experience

I worked with another designer to bring this to a medium fidelity keynote storyboard detailing the experience and proposed concepts that fall within in it. Defending the end to end approach against pressure to cut back to feature stories. I then presented this to the entire i2 team for Q3-Q4 scoping as well as in numerous user interviews where it was utilised as a tool for evaluative research. The query experience was chosen as a focal point, and following demonstrates a prototype of the (pre-visual) UX vision.

Highlight: Advanced Multi-source Query & Filtering Experience


A new UX effort is born. (Taking co-leadership)

 

We were still approaching this shortsightedly.

Even with our end to end experience this was still only looking at the immediate release cycles, building by capabilities from the existing product in isolation. We had no To-Be scenario. With concern around this approach, myself and another UX designer on the project started to design ‘an i2 without the constraints and baggage of what came before’. Focussing on the best user experience possible. All on the side, so we could measure success of each release not on improvement over the last release but on how it measures up to our perfect vision.

A new process & codename: 'EIA Great'

From this, we bought some time in an hour here and there from our lead to work on the new proposal. codenamed ‘EIA Great’.
Due to the success of that tiny amount of time over different locations even though the product pivoted away from the federal space for hallmark, the methodology and approach we used were adopted as the way to proceed by i2.  

Research Reviews

With those years of domain knowledge under i2’s belt and the new users and interviews already being pushed through, the first step was a deep analysis of existing research findings to collect and collate what was known and draw out the deep insights to back up any design action moving forwards when presented to the wider team.

Product architecture.jpg

On Rails Playthroughs

Sometimes, static wireframes alone can’t quite demonstrate the precise and delightful interactions and animations that can be the key to creating up a powerful experience - something I felt each quarter needed for morale.

For this highly detailed animated keynote run-throughs of a user’s journeys, like those in the early projects of design camp served extremely well to inspire the teams to work towards a common goal and level of quality.

Interaction Models

It gets hard to keep aligned around the user experience of a sprawling product such as this, and to keep the same logical model throughout. For the flip side of the equation; the detailed logical questions of the structure of our offering, I maintain a giant .ai file of every stage of the web client, detailing how to reach every other page to ensure there are no experiential gaps.

Radical Cross Location Collaboration

A major challenge stemmed from the majority of the design, development and management resource being situated in Cambridge, difficulties arise from being outside. To combat this, three of us instigated an extremely expedited set of meetings to align and work at a high pace, taking a room in the London Southbank studio and focusing on a deep dive into a single user experience each session. During these we would often create the outline of the work for the next few weeks.


Key Insight: The 90-10 balance.

The 90-10 balance refers to the fact that currently analysts spend 90% of their time conducting menial tasking and ‘hand cranking’ analysis, only leveraging our tools and analytics for less than 10% of their time.

This is time and time again raised as the clear and major pain of all analysts, and the greatest opportunity to deliver a truly signature experience. Concepts to address that are key.


Domain Specifics

In the world of intelligence analysis its important to take it one step further than industry verticals, through finding out both the Industry; Law enforcement, Federal, Military, Banking etc, it unlocks huge potential in narrowing down the scope and where to build accelerators. In this project we focussed on the U.S. federal space; opening up the need for powerful reporting and smarter views into great masses of hidden data without busying the surface.

Peripheral Insights

One of those smarter views that flourished was my concept of “Peripheral Insights”. In i2 typically each entity would have a large number of items connected to it, that you cannot see or know exist until you attempt to expand/filtered expand and receive possibly an overwhelming amount of results (a ‘death star”). But through the domain specifics set up, a single leading question can search a layer down and pull out and suggest results that might be of interest to the case, allowing the analyst peripheral vision into their data.

Insight Recommendation Engine

Today analysts often structure intricate queries to ‘ask the perfect question’, to try and find who might be their suspects. However in most cases they already have a good idea of lots of information on that suspect (possible name, description) as well as their context (committed violent crime in this address). Through deeply integrating IBM’s Identity Insights, we developed a way to fill out as much around a suspect as known, and then use that data and its context in the chart to pose the question of our datasources, bringing back the most likely matches.