Have a lot of CRM data, but can’t make heads or tails of what makes one customer buy and another not? Perhaps IBM’s “Watson” can help you solve the mystery. You might be surprised what the cognitive system will be able to tell you – and help turn “chaotic data into business decisions”.
On 29 July, IBM held their “Moving to Hybrid Cloud Data Management Deployment” event in Causeway Bay in Hong Kong.
The event consisted of several parts, starting off with Hybrid Cloud Data Management as a solution to common challenges for an IT person: “over budget projects”, “data warehouse capacities maxed out” and the problem of not being able to move “some analytic workloads to a public cloud due to the sensitivity of the data”. IBM presented their Hybrid Data Warehouse as a solution to some of these common problems.
At the event, IBM also gave an introduction to docker, dashDB and SoftLayer and presented a big data analytics case study with the goal to “accelerate the pace and agility of insight and innovation”. This by using one or more of the “fit-for-purpose databases” that are available, such as RethinkDB, MongoDB, PostgreSQL and “self-service analytics tools”, such as self-service BI (Business intelligence).
Watson’s take on CRM data
But what was the most interesting came in the 2nd part of the event, when Alan Fong from IBM showcased what IBM’s cognitive analytics? “Watson” can do with some pretty basic data from a CRM (Customer Relationship Management) system. The result was quite impressive.
At the start of the case study the presenter had:
- A long list of clients from a bank that had sold RMB (renminbi)
- A long list of clients from the same bank that has purchased CLDs (Currency Linked Deposits)
- An excerpt from the same bank’s CRM system where each client was defined by name, occupation, income, age, address, etc. (i.e. normal fields (columns) in a standard CRM system for service organization).
The question that the bank wanted to know was: What is the profile of the client that is most likely to purchase CLDs when they come into the bank to sell their RMB?
As Alan noted: “I know all the customers are selling RMB, that I know, but how can I detect those persons that are very likely to purchase the CLD?”
Alan then uploaded the CRM data into Watson and after a few seconds, Watson “presented” a heat map showing the different occupations of the customers and their likelihood to purchase a CLD when selling RMB. Alan noted that Watson had found: “a few occupations of the clients at the bank have a higher tendency to purchase CLDs when selling RMB”
As noted above, the data that was imported into Watson was CRM data in form of rows with names and then columns with different properties for each customer – one of them being “occupation”.
Watson was able to show the correlation of “selling RMB and purchasing CLD” behavior against different columns (fields) in the CRM system. This correlation was represented as a “spiral” diagram where 100% correlation was in the middle and the other levels of correlation between different columns (fields) in the CRM system was shown on the spiral arms. In this example, Alan said that the correlation between “selling / buying” and the “occupation” column (field) was 75% – which was the highest level of correlation among the different fields available in this CRM export / import.
Alan then went on to show how Watson could present the data in different formats and how you as a user could ask natural language questions to Watson to get more insights into the CRM data.
Let Watson make your sales more efficient
In a world of limited resources and limited time this information is of great value since you can focus your sales efforts where it would give the best? result.
Imagine the following:
- A bank teller meets a client that wants to sell his RMB.
- The bank teller has the option to suggest to this client that he should meet one of the bank’s CLD advisors.
- At the same time imagine that the bank as a limited number of such CLD advisors and that the bank wants to fill their schedules with potential clients that has the highest probability to purchase a CLD.
Now, based on having the above result from Watson, and some data of customer flows and number of clients per day that approach the tellers to sell RMB, you could as a bank calculate the most efficient way to allocate the CLD advisors’ time. Not on a “first come first serve basis” but by saving space in the CLD advisors schedule so that you have resources to handle the “persons that are very likely to purchase the CLD.”
What should be your next step?
If you have customer data that you can’t make heads or tails of, then Watson may be a good way forward. It’s easy to sign up for a trial account with IBM and stat to use the power of their services. Of course you need to make sure that you handle sensitive data with care, so don’t press “upload” before you talk to an expert in the field. But nevertheless, it’s clear that Watson can do a lot for any company that wants to turn their data into business decisions.
/Jens and Gushi
At the pen for Ripple Effect Consultancy
IBM Watson: http://www.ibm.com/watson/