For many organisations, investing in Business Intelligence (BI) as a capability is a decision that requires significant planning and investment. For anyone stuck in analysis paralysis, this PoV is designed to offer some basic criteria for identifying a use case that lends itself to getting value from your data today.
The promised land
Some people haven’t begun to collect data at all; others are struggling to make the business case for a team to setup, host and manage an enterprise-ready data cluster; others still are struggling to give employees the right tools to make something of the data they have already without needing to come to IT. In all cases people are struggling to realise the value of this data for their business, i.e. insight that will drive revenue, reduce operational costs, improve capital decisions etc. These benefits are the promised land.
“A journey of 1000 miles begins with a single step.” —Lao Tzu
Identifying the right use case for a PoC will help you to demonstrate the real business value of big data more quickly. How do you identify such a scenario?
The most important two variables to consider are:
a) value: an understanding of the target business outcome, who is responsible for this, and at least one hypothesis about what insight, at what time, and in what context will help achieve this outcome. This is important to help you evaluate to what extent the anticipated insight could deliver lasting value to the business.
Use case value scorecard
b) complexity: a shortlist of relevant data sources you will require or have available and the complexity of working with each one of these data sets. This is important to help you evaluate whether both the challenge of collecting the data and transforming it into a usable form are going to be within realistic expectations for a PoC.
Data complexity scorecard
If the total complexity of the data you are considering for your use case scores 5 or over, consider simplifying it or consider another use case instead.
Once you have a workable scope, follow a PoC approach to build, measure, learn and demonstrate results through insight using a self-service analytics platform such as Tableau, PowerBI or Qlik and make recommendations that deliver business value—if available.
Putting it into Practice—two examples
A logistics story
A logistics and warehousing company was struggling with unplanned downtime for its forklift machinery. It wanted to improve uptime to 98% to avoid late deliveries. The Warehouse Manager, Service Engineer, Staff and Machine Operators were jointly responsible for this outcome. This was determined to have a business value of 3 because of the target productivity uplift of 8% for their warehouse resources.
The team hypothesised that forklift telematics data could help signal when a fault was imminent and they wanted to validate this. They decided they needed the following data to make this PoC possible and scored each one using the complexity scorecard:
Forklift telematics data (2), breakdown reports (0), machine usage data (0)
This particular use case offered relatively high business value and low total data complexity (2), and was therefore a good candidate for the PoC.
After blending and analysing the data, the company was able to validate their original hypothesis and pinpoint the energy signature of the engine as a key indicator of imminent breakdown. Moreover, because of their ability to ask new questions and interact with the data, they were further able to identify that the fan belt was responsible for 78% of breakdowns, and that in all cases a breakdown occurred after four or more hours of continual use of the vehicle.
Based on an evaluation of the results and costs involved the team agreed that the solution could deliver ROI at scale and recommended the following changes:
Machine operators now have an app that notifies them in their heads-up display (e.g. Vuzix smart glasses) to dock their vehicles after 3.5 hours of use, or when the energy signature is indicative of a impending breakdown. Service engineers are also tasked with regularly
replacing fan belts on all vehicles as a preventative measure as part of their service routine.
A retail story
A leading retailer of shoes and apparel was faced with lower than anticipated turnover and wanted to increase the profitability of its stores in key locations. Store managers were being held accountable for better results and IT wanted to empower them to succeed.
Representatives from IT and Store Management jointly hypothesised that product placement was a significant sales influencer and constantly changing. They wanted to validate this, so that store managers nationwide could optimise this on a daily basis. Together they identified the following data sources and scored each one using the complexity scorecard:
Sales data (0), product placement data (5), store data (0), weather (0), discounts (0), celebrity influence (8), regional demographics (0), marketing campaign data (2), audience opinion data (reviews, social media etc) (8), macro-economic trends (0), regional economic trends (0)
This particular use case offered relatively high business value but the data needs were complex so the company had to simplify the scope of the PoC in order to make it achievable. They decided to drop a number of data sources and focus on a) validating the core relationship and b) placement of shoes rather than apparel as well. Hence the revised scorecard:
Sales data (0), product placement data (4), store data (0), weather (0), Discounts (0), celebrity influence (8), regional demographics (0), marketing campaign data (2), audience opinion data (reviews, social media etc) (8), macro-economic trends (0), regional economic trends (0)
In order to overcome the challenges of data capture, the company decided to host a “data warehouse” with Google BigQuery, and successfully developed a proprietary method of recording product placement throughout the store and feeding it to BigQuery.
After blending and analysing the data, the team was able to establish a statistically significant relationship between product placement of shoes and turnover for equivalent stores. It was agreed that there was significant potential for communicating recommendations to store managers daily so that they could take steps to optimise placement. However, the team also agreed that the solution was not ready to deliver ROI and that they should iterate their PoC further, including other data sources to nuance the relationship (with celebrity influencer data for example), and better matched stores across their portfolio for improved recommendations.
We hope that this approach helps you to simplify the conversation about getting started, and that these stories demonstrate the possibility of realising the value of big data within your organisation without making considerable financial investment first.
Mubaloo is a specialist mobile partner, consulting on business transformation through mobile and the delivery of bespoke, enterprise-grade applications.
Mubaloo has a wealth of cross-industry expertise, which comes from experience working across sectors developing innovative mobile strategies and intelligent mobile apps. Mubaloo helps to transform business processes, productivity and customer engagement, all through the use of mobile.