November 25, 2014

Data Science + Behavioral Science = Force Multiplier ! 3 Real Life Stories

The word “Force Multiplier,” in military usage, refers to an enabler or a combination of enablers, which make a given force more effective than that same force would be without it.  
So what force multipliers can be used in the world of Data Products ?
Based on our experience in curating intelligent data products for the industrial world we feel a Data Product needs a healthy dose of 2 things
1. Data Science = Seeing the unseen       ( Surface actionable signals previously undetected to bake into a data product )
2. Behavioral Science = Making it relevant      ( Making data product relevant to the daily context of tasks a person does to accomplish concrete business outcomes )
While there has been a lot of conversations around data science, we at Flutura felt that it was important to balance this with conversations around behavioral science which in our opinion is a force multiplier in driving adoption of data products.Having created  intelligence platforms for the industrial world, we have our own share of successes and failures. One key learning from our failed data products was the need to weave behavioral science into data products. Allow me to share 3 real life stories from the trenches which shaped our thought process.
Real Life Story-1 : Using gamification to reduce alert Turn around time ( TAT) & operational risk in Connected Buildings
Flutura was solving the problem of mitigating operational risk as a part of a Smart city project ingesting billions of events and out of condition alerts from boilers, chillers, fire sensors, smoke sensors etc across 110,000 buildings in real time . These streaming events were monitored by our Cerebra data product from a command center manned by 40+ command center operators. Monitoring exceptions in a command center and responding to them in real time. How do we heighten engagement of the console. By creating a gamified experience where operators can level up with respect to other operators with regard to TAT ( Turn around time ) sufficiently increased engagement and minimized operational risk from unattended alerts
Real Life Story-2 : Using Social Proof to reduce Microgrid Instability
The other real life experience involved solving the problem of regularizing prosumer behavior in the Microgrid space in utilities. As solar and wind power gets integrated into the complex utility grid its important for prosumers ( folks who produce energy + consumer energy) to contribute back to the grid.So the problem we were trying to solve was how do we decrease grid instability by increasing prosumer stability ? Again an energy data product which benchmarked the prosumers behavior with neighborhood behavior helped in heightening the awareness in a humane way and achieving the grid stability outcome.
Real Life Story-3Weaving a data product into a daily ritual
In industrial context many assets are in the field with default factory calibration. Exposure to field conditions necessitate re-calibration . Flutura created an asset re-calibration reccomender ( data product) into woven into the existing ticketing system by creating an automated ticket which can drive the front line action as opposed to making the operator open a new application or dashboard to use. Weaving a data product into the fabric  of a daily ritual which is routinely executed instead of making data product consumers  alter their daily routine. It is now a part of a daily journey which is already undertaken as opposed to forming a new journey for the data product consumer 
Why is Behavioral Science important to Data Products ?
1. It humanizes data.
It puts Big/Small data, Structured/Unstructured data,  Low velocity /High Velocity ,  Machine learning, Map reduce, Collaborative filtering, Text mining, Graph theory, Apriori analysis and all the innovations which has enabled us to solve technical challenges  and neatly  bundles them it into a nice "box" which solves a real world business problem.
2. It aligns mental models ! Enhances "Consumer Resonance Index"
Consumers of data products think differently from the creators of data products ! Its a simple fact but as with many things simple, often overlooked.  Incorporating learning's from behavioral science can heighten the resonance the data product has with real front line users which impacts a business outcome. ( CRI- Consumer Resonance Index as we at Flutura call it ) 
3. It makes the data invisible !
Data fades into the background and jobs which need to get done comes into the foregroundThe best data products make the data invisible and the tasks visible ! It really should not "smell" of algorithms or data. Behavioral science understands these subtle nuances and the neural pathways which need to get activated in order to drive repeat engagement with the data product and get daily jobs done !
So how do we think about this ? 5 Conversation starters
1. HUMANIZING - How do we humanize a data product to heighten its engagement ?
2. RITUALS - Are we hooking the data product to daily "ritual" being performed by the data product consumer ?
3. IMPACT VISIBILITY - Are we amplifying the visibility of the outcome achieved based on the action signals sent by the data product ?
4. CONSUMER RESONANCE INDEX - How is the data product engaging with the hard wired mental models which are embedded deep down ?
5. CROSS TRAIN DATA SCIENTIST - Do we need to cross train our existing data scientists in best practices from behavioral science to reduce the risk of non adoption ?
As our friends in the military from West Point would say, "We need more force multipliers".
Data Science + Behavioural Science = Force Multiplier
When used in combination, these techniques deliver greater value than when used in isolation.

For all of us at Flutura, the learning that Behavioral Science is as important as Data Science in our arsenal to create impactful data products was very liberating . This was an important moment of truth in our quest to creating engaging industrial intelligence platforms. We hope you can learn from our mistakes and  good luck with your efforts to create HUMANE DATA PRODUCTS !

November 9, 2014

Why Data Scientists create poor Data Products ? 5 humbling lessons

As the consumer and industrial world gets massively digitized Data products are being baked into critical processes at a very high rate. These data products distill signals from massive torrent of human generated and machine generated data to drive a front line action . At this point we wanted to distinguish between 2 types of data products which we have seen in the market place
1) Consumer Data Products : Data products created to harness human generated data intelligence like Sentiment Analyzer, Reccomender engines , Social graph analyzers, Digital Purchase intent detector
2) Industrial Data Products : Data products created to harness machine/sensor generated data intelligence in Industrial IOT world like Asset reccomenders, Mean time between failure calculators etc
In the consumer world, Data Scientists were able to create an amazing job of curating game changing data products primarily because they were able to relate to the consumer context be it the decoding digital intent from a sales funnel or suggesting the next best action to a digitally engaged user
In the industrial world we have seen its relatively difficult for pure play data scientists to relate to the machine world and as a result a lot of data products which have been created with the best of intentions have failed to make it to the operational side primarily because of the dissonance in mental models between an industrial engineer and a data scientist. So what can one do to increase the chances of Industrial data products being adopted in the engineering world ?Based on Fluturas experience in curating Industrial Data Products here are our 5 mantras.
Learning-1 : Be Engineering backward, Instead of Data Forward
Data scientists tend to get seduced by the algorithms and the platforms processing billions of event data. In the process they lose sight of the problem to solve.For example consider making an electrical or mechanical engineer the product manager . He/She would stay focused on the engineering problem to solve. Its easier for an engineer to learn data science than for a data scientist to learn engineering nuances
Learning-2 : "Walk a mile in the Engineers shoe"
Its very important for data scientist to empathize with the conditions under which a front line engineer would engage with an industrial data product. A data scientist creating a data product need to have a healthy appreciation for the tasks which happen before and after a data product is consumed and the conditions under which they are consumed.
Learning-3 : Industrial Engineers quality threshold > Data Scientist quality threshold
When an engineer makes a product and releases it, it has been rigorously tested before being launched so there is a certain quality expected. Flutura experience in the industrial world shows an Industrial engineer has higher expectations of quality from a data product than a data scientist. Even a minor software bug makes the industrial engineer more crazy then a software engineer.This difference in expectations needs to be harmoniosly managed
Learning-4 : Analysis is not a job to be done
Front line Industrial engineers are paid to take action. For them analyzing is not a real job to be done - plain and simple. So any data products which aligns to that mental model resonates. For example in the smart grid scenario a data product which recommends a specific asset to stabilize the grid is worth more than a barrage of pretty graphs on power quality. Value perceived by an industrial customers lies in the ultra specific actions and not on the compelling visuals
Learning-5 : Aim for both heart and mind
A data product is being used by a human being. Its very important to align the data product user experience to evoke the right feelings. Hypersensitivity to this important feature can make the difference between success or failure in industrial data product adoption
To conclude, interesting things are always at the intersect !
As the industrial world gets increasingly digitized, these 5 learning would go a long way in managing the fusion of 2 different worlds which are colliding more often :). Brickbats ? Thoughts ? Comments ?