When someone tells me they want to do analytics, I say that it is easy. I can explain it in five minutes. However, if someone says that they want to transform an organization using analytics, my reaction is, “How long do you have?”
It is not easy. Let’s dissect how to do it.
There is a new buzzword in the block: digital business. What is that?
The idea stems from few businesses that managed to rethink the world.
Uber built a taxi company without owning cars or employing drivers. Airbnb built a hospitality service without owning rooms or hotels. Media and retail (i.e. Amazon, eBay) have been transformed by digital.
We call such a business a digital business. Digital businesses use digital technologies to fundamentally rethink the way businesses are done. The path to becoming such a business is called digital transformation.
Following are three good references. Each has a slightly different interpretation of “digital transformation.” At the same time, there is enough census to make it useful.
Why do we need this? New technology has amused us and connected us. Yet, not all are amused. A 2012 MIT Technology Review headline ran You promised me Mars colonies. Instead, I got Facebook. Many have argued that despite amusement, new technologies have not lead to fundamental advancements.
Technologies like the internet are omnipresent. However, critics argue that those technologies have failed to fundamentally change our lives or add major efficiencies like advancements did in the Industrial Revolution.
My hope is that redefining around the digital can bring in those missing efficiencies and leapfrog us to the future. It's needless to say that if any organization can do that, they will leave their competition in the dust. Hence, it is not much of a choice. Any organization who wants to be around in 10-20 years will have to do that.
Companies like Uber, if they do it right, can change the way we live. For example, Uber provides a future where many of us can survive without a car, reducing traffic and reducing costs.
Analytics in Digital Business
Three classes of digital technologies can play a part in this transformation: analytics and AI, social mobile and IoT, and crowdsourcing. This post explores the role of analytics and AI.
The following picture shows a component of a digital business. This is by no means the only representation. Yet, the following picture captures most ideas discussed under digital business. Furthermore, it lists five areas where analytics can play a major role within a business.
The next sections explore each in detail.
The key idea is to collect data about the organization and use it to improve operations. This is the most widely discussed benefit of analytics. The unstated assumption is that the organization has a lot of friction and inefficiencies. I am sure most of you will agree that this is true.
There are many use cases that fall under this. Following are a few.
Predictive maintenance i.e. production lines, equipment, vehicle fleets, and sites.
Optimizations, scheduling, and responding to issues i.e. making sure the right person is playing the right role and has access to everything they need.
Fraud detection and prevention.
HR analytics i.e. finding the best candidates, filtering candidates, performance appraisal analytics and proactive intervention, andchurn prediction.
Security and surveillance.
Key for optimizations is knowing what to measure. We used the term KPI (Key Performance Indicators) to describe those measurements. KPI is a simple indicator that represents an important aspect of the situation.
A good example of a KPI is thinking about canaries in the coal mines. Years ago, miners took canaries and other small birds into coal mines. These birds are very sensitive to oxygen levels. If the oxygen levels are low, they will get knocked out. Then, humans have an advance warning to rush out of the mine.
Following are other examples of KPIs.
- GDP and life expectancy for a country.
- Revenue per square foot in a retail store, revenue per employee, and revenue in the organization.
- Customer lifetime value (CLV) and customer acquisition cost (CAC) in sales and marketing.
Yet, the given KPI will only see a part of the picture. Hence, a narrow focus on the KPI can lead to suboptimal outputs. To tackle this problem, Andy Grove (Intel) argued in his book High Throughput Management that KPIs should be used in pairs. The first KPI should measure the output (i.e. processed claims count) and the second should measure the quality (i.e. mistakes occurred). Together, two KPIs provide a holistic view of an organization.
Most domains have well-defined KPIs. Defining new KPIs is hard work. Before optimizations, you must define the KPIs. Then, you need to either measure them or find a way to approximate the KPIs. What you need to measure is not what is easy to measure, but what is useful.
Facts do not cease to exist because they are ignored. — Aldous Huxley
When you have found the right KPIs, the rest is relatively easy. We can use KPIs to find problems, come up with and apply solutions, and to find the effectiveness of our solutions. We can apply the process in an ongoing basis to improve operations.
Get Close to Your Customers
Analytics can optimize organization’s interaction with customers and create new dynamics. This can happen in many forms.
Analyzing customer-product interactions can improve the functionality and user experience of a product (i.e. adding a timer to the coffee machine).
Send customers periodic emails with key usage patterns and insights. This will also provide opportunities for cross-selling and upselling.
An app connected with the product provides a communication channel to the customer. For example, the customer can use the app to report any issues, provide feedback, and schedule appointments. At the same time, you can use the app to keep then informed, push offers and advertisements.
Connect with the customer via social media and follow them to understand their likes, dislikes, and sentiments. These insights can be used to take product decisions. For example, if customers use some other product with your product, maybe you need to partner with another company or create a competitive product. Furthermore, the same data can be used to track sentiments and understand the brand’s ups and downs.
Find problems and take the initiative to solve them before other customers experience the problem and push updates to other customers proactively.
Keeping touch with customers regularly will let you run efficient marketing campaigns. For example, we can do events and activities in the areas with high density of the customers. This creates network effects that lead to more sales. For example, we can invite a few customers when you make a donation to the hospital or sponsor a school event.
Create users groups to let the customers talk to each other and learn from each other. You will find network effect drive them to more adoption. Mercedes-Benz Club is an example of this idea. However, in the world of social media, these clubs can operate cheaper than before.
Improve Marketing and Sales
Marketing is scouts and sales are hunters. Customers often start as visitors on the website. Marketing brings in new visitors in several ways.
Creating content that perspective customers will be interested in and publishing them on the site.
Engaging in lead generating activities such as workshops, sponsoring events, external publications, etc.
TV, SMS, and print media advertisements.
Then, marketing identifies, keeps track of, and nurtures visitors through newsletters and other activities. Sales engage in making the sale when ready. Analytics can identify and track leads as they move through the marketing and sales pipelines.
Following are some things analytics can do.
Optimize the time spent on lead nurturing by ranking leads by their effectiveness (i.e. using Machine Learning models using historical data).
Track, analyze, and understand the effectiveness of published content, traffic sources, and the impact of the message on traffic.
Running an efficient digital media campaign.
Mapping out the customer journey using the data collected from their interactions enables us to understand customers and to guide them through the journey. This includes recommending the best material (i.e. other customers from the same domain and others who do similar things) and helping get them through the sales process as fast as possible, as well as perspective customer profiles to get them to the right customers to build confidence.
Understand the physical location of customers and use that to optimize the one-to-one interactions.
Tracking topics related to product and producing content that utilizes those topics.
Create New Revenue Streams and Products
Under this topic, there is no limit to things you can do. Following are some favorites.
Some products need humans to tune them, and it is hard to get them right. Machine Learning can take away the burden. For example:
- An oven that decides how long it takes to cook and adjust temperatures automatically.
- A washing machine that determines the size of the cycle and optimizes the water.
- Smart products that optimize for energy bills and comfort (i.e.. Google Next).
- Bose headphones that let you talk in a noisy environment.
- Fit gadgets i.e. Fitbits, shoes, etc.
- Cameras that use multiple small cameras and combine images using Machine Learning to produce high-quality photos.
- Let you control or check the status of a product remotely (did I turn off the oven, heater, iron?).
- Products that provide new input methods (i.e. take a picture instead of typing or speaking).
Solve a Problem
- A fridge that keeps an inventory of what is inside.
- Tracking things you forget or lose (i.e. key, wallet).
- Surveillance (i.e. when is the last time something moved?).
- A medicine dispenser that reminds you to take medicine.
- A cup that measures content (i.e. sugar intake).
- Trackable wallet, backpack, etc.
- Digital twin for analysis, reasoning, and diagnosis.
Create Value Networks
When you run a business, you create a network of many different people and entities. It is likely that this network can play a bigger role that part of the business and used to your advantage. This network that connects customers, the organization, supply chain, and distribution chain is called a value network.
For example, Walmart provides detailed analytics about their products to suppliers, which let them improve products. These analytics act as value added advantage for the supplier for working with Walmart. The same idea can be used across many businesses where you can not only use intelligence but also pass over intelligence to other parties.
A company value network might provide a great opportunity to create and distribute a new product. A good example is the Nespresso Case study, where Nestle used their distribution chain to sell expresso machines and how they managed to defend the revenue stream through the coffee pods.
Moreover, there may be other products that sell together with your products. For example, many items go together such as diapers and wet wipes or pizza crust and cheese. Understanding them might give you several opportunities.
You can compete by creating a new product and using your distribution line to sell it.
Make a deal with the supplier of the other product to either share revenue for sales you bring to them or to share the marketing costs.
Negotiate with the supplier to give your customers a better deal.
Finally, the information that can be collected through the value chain might be useful. For example, the distribution chain can provide data about the demographics of users or their reception.
It is easy to do analytics. However, it is hard to change organizations and change the world using analytics. The idea of digital business brings this problem into focus.
This post tries to look at organization holistically and discusses how analytics can be used to rethink how to do business. This post intentionally does not discuss how to do it (technically), rather tries to cover what to do using some use cases.
If you are looking to implement these use cases, you have a choice of many tools. Among them are WS02 Stream Processor, Apache Spark, Apache Flink, Apache Storm, and Kafka Streams. Please report to respective web sites for more information.
I hope this was useful. If you have any thoughts or think of new use cases, I'd love to hear them though the comments.