OpenIDEO is an open innovation platform. Join our global community to solve big challenges for social good. Sign up, Login or Learn more

[UPDATED 2/20] Reframing the question: maximize "endorphin-per-dollar" as a core "job to be done" of financial firms

Not more money: more happiness per dollar. People are happy to pay for AI-powered financial advice that makes their life good.

Photo of Gianni Giacomelli

Written by

Who is your idea designed for and how does it support the dreams and obligations of those 50 and older?

While the idea could work for anyone who uses financial resources, people above 50 will benefit more from it as their financial resources will become constrained at some point. The ability to help them direct their spend to what makes their life better will reduce "financial waste" i.e. Money that is spent on things that don't make their life happier.

The inspiration for this idea is simple. Financial institution's core "job to be done" should be reframed away from "fee per transaction" (which made sense in a low-data, transaction-is-hard world) to "fee per life outcome". Financial firms have or can get access to data (both existing and new) that shows inferences between lifestyle choices, saving/expenditure, and happiness of households. Maximize "life is good" (LIG). Mining that information can power an artificial intelligence powered "virtual assistant" that provides tips to reduce the financial expenditure while preserving quality of life. Financial institutions can monetize that service, and make it increase the "stickiness" of their service. 

Consider the following story.

Two 50+ year old people have 100K $ per year disposable household income each. Each of them has the same financial commitments - kid's school, mortgage, healthcare. One is much happier than the other - not because who she is, but how she spends her money. An analysis of her saving and spending patterns (obtained from credit card transactions etc), as well as other key indicators such as social relatedness (social media friend behavior, sentiment), may reveal that she makes financial decisions that cut "financial waste". The more data the virtual assistant - call it "Happy Penny" - will gather (from bigger communities of subscribers, or from new data sets eg through continuous polling of individuals asking specific questions e.g. "Life is Good" (LIG) levels, satisfaction with specific family outcomes, "the best purchase I have made in a long time..." etc), the more the tips will be accurate and actionable. Apart from the tips, Happy Penny's power also comes from the value of "nudging" people into action, reminding them of things that will make them happy. Artificial intelligence (likely machine learning that tests outcomes of "natural experiments") can be used for both identification of the right tips as well as pacing of the "nudges". 


Imagine this conversation between Frank, a retired, married empty nester, with Penny, his bank's "happiness coach" bot. Frank whispers to Amazon echo dot:

"Alexa, get me Penny"

"Hey Frank, good news - this month you seem to have spent 15% money for the things that make you happy. Do you want to know more?"

"Yes Penny go ahead"

"This month your money-well-spent score was at 73%, up from 65% last month"
"You have spent more money on restaurants - you seem to have invited people there, and you like restaurants and friends" "you also seem to have gone with some friends to the local cinema a couple of time" "and you have donated to the local pet shelter" "these are all things that show you're spending on things that matter to you"

"you've also saved a little bit of money on energy consumption compared to last year, which gives you a little to put aside" "Would you want me to suggest ways to use your resources that may make you feel good?"

"Yes please Penny, tell me"

"Frank, have you considered inviting a few more friends home for dinner or brunch a couple of times in the coming month? You can even have someone help you with the preparation, as you and your spouse may be busy. Alexa can help you with searching for one" "or you could go to the upcoming jazz festival in town, since you like jazz". "Actually, you don't seem to have spent much on music recently, and you may like some live concerts?" "Also, have you considered going to get some massage? You haven't in a while".

"Frank, do you want to know what I have also learnt this month from people similar to you, whose money-well-spent index is higher than yours?"

"Of course"

"People who have a similar wealth and commitments and have similar interests to yours, tend to be very happy when they spend one-third more money on their family and friends activities. They also seem to spend more money on short vacations that build memories, especially shared memories with others. Their investment profile is also slightly different than yours - they have a little more in fixed income at this point in time"

"And to wrap this up, I have listed these three things that made people like you really happy last month - some of them may be new. Here you go. One is the florist. The second one was the skating rink in Bryant Park. The third was the subscription to streaming music on the internet"

How did this conversation happen?

1. Frank had compiled a comprehensive "what makes me happy" questionnaire (whose results stay between Frank and his bank, and are never sold to anyone)
2. He had started using the Happy Penny app, helping it cataloging the expenditures with 15 minutes of time a month (i.e. Disambiguating the classification of expenditures that Happy Penny's AI wasn't able to firmly classify e.g. Dinners that could have been business related etc). To make Penny even smarter, Frank flags expenditures that made him (and/or spouse) really happy. All of this must be done through a gamified UI
3. Happy Penny's algorithm compares his spending patterns to people similar to him who are happier, and continuously suggests changes to the expenditures. We can complement that corpus of knowledge with research from behavioral psychologists, and offer those tips as part of Penny's input
4. Over time, and as the client trusts the bot more, we could consider introducing additional behavioral questions (e.g. How many good friends did you meet this month? And family members you like?)


Local communities can likely act as a force multiplier - it is likely that in-person connections quality are comparatively powerful contributor to satisfaction. Imagine the role of the Community banks and credit unions as a driver (and possibly a contributor of funds) for action for community level organizations, such as church, sport, schools, and other groups (art, culture, environment, etc). They can inform action, and encourage individuals to get involved in the very activities that may generate the maximum happiness return on money. 

It is possible that the virtual assistant will benefit from the partnership with other organizations e.g. Healthcare that own significant data (e.g. Depression outcomes and their drivers). 


See the attached pdf for more detail.

In summary: we need to create an algorithm that infers what specific financial outlays (the gym subscription, the florist, a dinner with good friends, an outing with the grandchildren, watching the favorite baseball game, paying for piano lessons for them, taking some courses, donating and volunteering to a community service, or spending on some acupuncture for that rheumatic ache...) makes people feel that "life is good". Each person feels differently about the value of that money, and the suggestions should be based on (a) what typically has made them happy (but they aren't doing all that much) and (b) what makes "people like them" feel great (there are likely a bunch of good ideas that can be crowdsourced by the assistant). The objective is to build a virtual assistant, a coach, that nudges people in action. 

What's important to recognize is that there is sometime a difference between what people STATE that makes them happy, and what really DOES. In the algorithm, we should treat differently the items that fall in one of the two categories but not in both (see page 2 and 4). 

What early, lightweight experiment might you try out in your own community to find out if the idea will meet your expectations?

Review of literature around life outcome satisfaction and expenditures. Review of ethnographic studies of communities with strong self reported happiness levels, and understanding what expenditures they make. Do a small sample, low tech survey of people in similar financial conditions and identify drivers of differential life satisfaction level - and test the acceptance of "people like you seem to benefit from XYz"

What skills, input or guidance from the OpenIDEO community would be most helpful in building out or refining your idea?

Thorough understanding of elderly well-being drivers and their interplay with financial choices. Machine learning experts. Gamification experts.

Tell us about your work experience:

I am the chief innovation officer of a digitally enabled professional services firm that spun out of GE.

This idea emerged from

  • An Individual

How would you describe this idea while in an elevator with someone?

What is a community bank to someone aged 50+? What is should and could be, is a provider of ideas on how to spend and make those dollars feel your life is good. By understanding personal profiles based on spending patterns, and correlating them with life satisfaction levels across large samples, a community bank can have a powerful tool to advise clients, and increase loyalty. This is Happy Penny.

How might your idea be transferable to a large number of people?

Happy Penny benefits from large data sets. Its operating model is exponential - more data, better prediction, more engagement, more data. At low scale is the most critical aspect - even in the initial phase, when algo doesn't have much data, the recommendations should feel great, so that users engage frequently and share with their friends, relatives etc. That can be achieved by using conventional questionnaire analysis, supplemented by basic social media analysis, and psychologists advice.

How do you plan to measure the impact of your idea?

At scale, the metrics are pretty obvious (engagement, retention, movement of the Life is Good scores). At smaller scale, some of the metrics for start up apps could be applied. We should also consider the impact on brand image for the bank, as the halo effect can be significant.

What are your immediate next steps after the challenge?

A MVP could be built pretty easily.

Attachments (1)


Join the conversation:

Photo of Dave Krimm

The psychographic segment to which this app might appeal would seem to me to be quite small: are there really that many individuals who are uncertain regarding which marginal expenditure would make them happiest? And it's not clear how a financial institution might "monetize" utilization of this app: it's not an investment product, or a cash management product. So the ROI on an investment in this app would appear to be quite uncertain.

View all comments