The DataHero Blog

How a Dark-Horse Prototype Improved Sharing

March 10th, 2015

dark-horse-banner

At DataHero, we believe that prototyping and testing will lead us to a superior product.  What might not be clear is that prototyping doesn’t always equal iteration; specifically, a prototype doesn’t always mean testing a possible final solution.  Sometimes it means creating a long-shot design that’s only purpose is to inform a later design.  This concept and philosophy couldn’t have been more true as we developed DataHero’s sharing functionality by creating a dark-horse prototype.

The Problem

Our process to determine what to build started the same as other features at DataHero – with user interaction and research.  I tried to interview at least 2-3 users every day and focus my questions and interactions across our defined user-personas as well as level of experience of those using DataHero.  To my point above, the ranges of how people expected to share came in many different variations.

There were so many options on how to create sharing functionality within a visualization product that user feedback almost made the project definition worse.  We always start with users first at DataHero, but the requests and ranges of expectations for sharing were dizzying.  For example, solutions and user requests ranged from fully-collaborative exploration, to controlled view only chart sharing, to sharing charts embedded in email, to the ability to embed charts in any website.  With sharing, we were going to have to try something radical and different to succeed.

The Path To a Solution

During our initial research, one of our findings was something we didn’t really expect: people wanted the ability to fully collaborate on data analysis with colleagues in an interactive manner.  This was interesting because the idea was similar to a Google Docs for data.  How do you let anyone access a dataset, change charts or analysis created on it and generally work in parallel together?  To our knowledge, this is pretty unique compared to any other data product on the market.  Letting other people change your chart, adjust your findings and basically let anyone on a team edit and modify any dataset was an extreme idea.  How would people react?  What features would be required?  How many people would use it?  Is this really something people wanted or were they projecting a sharing ideal?

One thing I learned quickly in Stanford’s d.school is that prototypes are a great way to better understand reactions to a solution.  The faster you can create something to really test your hypothesis, the more you will learn.

At DataHero, prototypes show us two things:

  • Things that need more work
  • Things that don’t work at all

[bctt tweet=” Prototypes show us two things: things that need more work & things that don’t work at all”]

The goal of any prototype is to really test a hypothesis.  It should be a step toward drastically different choices and not just an a/b optimization test.  These kind of tests and prototypes yield the best insights – and usually lead to another prototype to test something new.

The Dark-Horse Prototype

Because we had gotten such varying feedback, we decided to test the most extreme version of sharing we could in a dark-horse prototype.  A dark-horse prototype is a prototype that is meant to expand the design solutions by testing something that is “out there” and you don’t expect to work.  The rationale being that you will gain insight that will make your final design better, and you never know when something so crazy just might be the winner.

[bctt tweet=”A dark-horse prototype is meant to expand the design solutions and gain unique insights.”]

So we went crazy and created a prototype we knew would be our dark-horse. Our first chart sharing prototype allowed any person to share the root dataset with anyone else they desired; those people in turn had full access to create charts, edit the data types, modify charts others had created, and make their own dashboards based off this data.

dark-horse chart sharing prototype

Overall, we learned some valuable insights from this crazy prototype:

  • There are some organizations where this fully collaborative model is exactly what they need.  However, there is a lot of complexity trying to do version management, access level requirements and an overall level of trust required in colleagues and collaborators for this model to work successfully.
  • We learned that for most people “share” was more synonymous with “convince” than anything else.  They only wanted to share a view-only chart but not allow people to adjust it.  There is a lot of fear that people will “misinterpret” the analysis or data.  Overall, they want their data to tell a story.
  • We learned there are some attributes people want to allow others to adjust.  For example, changing time formats or groupings, adjusting certain categories displayed, etc.
  • In an organization, typically there are a few creators of analysis, but the majority of people are consumers.  They have no desire to adjust or modify the results, but rather just wish to be informed.
  • People expect to be able to share information in a variety of formats with their colleagues.  The easier it is to send an image in Slack, share a live link in email or export a PDF board ready packet, the more people will love your data analysis product.

The Final Result

DataHero Chart Sharing Overview

From our extreme findings, we proudly released our first pass of chart sharing and dashboard sharing in DataHero.  Our solution began with the most common users of DataHero.  For these users, we wanted to allow any user the ability to share charts with as many people as they wanted.  Simply press the share button on any chart to email or embed a link to a chart to be accessed by anyone who has the link anywhere.  If you use our automatic updates functionality to keep your charts up to date with the latest data, your shared link automatically shows the latest data.

The other use case was a team or group of users who wanted to share a group of charts in a dashboard.  DataHero is great at combining all of your reporting in one place, and across services, so sharing dashboards is a great way to keep anyone in your company up to date.  Because we know that many people will merely consume this data, we also created the all new dashboard view-only license to accommodate the largest of teams.

As you can see, creating a dark-horse prototype allowed us to test the limits, and we were able to create a variety of sharing options through the insights we uncovered.  Ultimately, our final solution addressed anyone from the casual user to the largest and most complex organizations.  We hope our new sharing functionality accelerates your reporting needs, and as always, we welcome feedback on how to adjust or improve it.

 

By Jeff Zabel

Create my Free DataHero Account

Get the fastest, easiest way to understand your data today.