Einstein meets Watson - a Bluewolf perspective

Profile picture for user mbanks By Martin Banks May 23, 2017
Summary:
As IBM’s lead on Salesforce delivery, recent acquisition Bluewolf is now well-placed to exploit the use of the Salesforce AI tool, Einstein, and IBM’s cognitive analytics tool, Watson, working as a pair.

einstein2
One of the cameos of the recent London appearance of the Salesforce World Tour was a potted use case that gave an idea for how Einstein, the  Salesforce AI offering, can work with IBM’s cognitive, deductive analytics tool, Watson.

The story line was straight forward – an insurance company has an information feed service from Watson about factors that can influence the number and types of claims that can be made against insurance policies. A classic example is the weather. IBM, having acquired The Weather Company last year to be a key data source for Watson to work with, is well-placed to analyse out predictions of weather events that affect specific types of business, such as insurance.

So an information feed that says, 'This type of nasty weather will occur in this area at this time, with this level of probability’ gives an insurance company, using Einstein to analyse its customers by location and policy type, a great line in customer service. It can advise that event is coming, together with advice on precautions that need to be taken before the event.

The customers know to take precautions to reduce their possible risk. The insurance company reduced its risk of having to pay out against the policy; and it had now given itself the perfect escape clause of having informed the customer of its risk. Miss out on executing even one syllable’s worth of the advice and the policy could be invalidated.

This coupling brings together the ability to work with specific, fine grain data about customers that Salesforce generates as a matter of course, and sets it against what might inelegantly be referred to as the bulk data of the world, analysed down and to the point where it is specifically relevant to a subject type or a defined segment of business.

This is a relationship that now sits at the heart of Bluewolf, a 15-year plus specialist Salesforce consultancy which, towards the end of last year, was snapped up by IBM. It is now, according to the company’s General Manager in Europe, Glen Stoffel, who started the original Salesforce consultancy programme for the company, part of IBM iX (its Interactive Experience Division) which describes itself as 'an integrated solution for design, business strategy, mobile, systems integration, and technological implementation’.

In other words, it combines consultancy with the `Do It For You’ capabilities of an SI in business areas where large scale analytics, AI and cognitive-driven decision making are key components of a solution, says Stoffel:

As far as CRM is concerned, we are where the rubber meets the road in terms of executing that vision. I make sure they combine the vision and messaging to the tactics and execution of CRM.

Inside out and outside in

The interface between Watson and Einstein is where customer specific data meets event or area specific information on a global scale. As an example Stoffel uses the notion of a financial adviser business, where Einstein is working on customer specific data, such as: 'Seven of 200 clients have suffered negative incidents on their accounts and they are due a quarterly review – call them’. He explains:

When they are called, the advisor is supposed to be an expert on that customer’s investment areas, and there are likely to have been 200,000 articles written that day on subjects relevant to the customer’s portfolio. Watson can analyse that data to deduce the investment trends and best advice, based on Einstein’s information feed on what investments the customer actually holds. That way, highly relevant investment advice can be given. This is about augmenting the ability of the humans involved.

This plays to one of the key capabilities of good sales staff in that they can spot when a customer is at risk, and given them the information tools that can identify the customers, the risks they face, and the tools they need to either avert or ameliorate it.

Sometimes, of course, this now means that Watson is the first line of customer interaction, where it is used to answer the simple early questions that customers can ask. For many of these it is no longer necessary for human intervention as the answers are pretty standardised. But it is an area where some of Watson’s additional capabilities, such as sentiment analysis, can be deployed to evaluate the way an interaction is progressing. This can then include a handover process to a human operator who has been fully informed on the customer, the query, and way the interaction has progressed.

Stoffel sees the Internet of Things as another major use case for the combination of Watson and Salesforce, but his biggest job for now is helping all those customers that come to an event such as the Salesforce World Tour in London last week and ask the straight forward question - where do I start with all this? And one of the key questions here is that many of them want to answer another very straight forward question - how much money is Salesforce making for us.

Here, he suggests, the combination of cognitive services and CRM can begin to prove the efficacy of CRM financially. Questions such as `How many customers did I save that were thinking of leaving?’ are among the narratives that he sees as important right now.

Overall,  he sees the combination of Salesforce and Watson as a tool combo for improving productivity, and that means starting with a strategy: for example having a goal of becoming a $20 billion retail company by increasing sales to existing customers:

So that will probably mean more SKUs per store, which in turn means more cross-selling. The next question is then whether that means working with the inside guys, outside guys or perhaps customer service agents. Suddenly there are many different actors that may play a role in such a cross-sell process.

This then prompts another important question -  how much is it worth to the business to solve such problems? In Stoffel's view this – a simple some of cost of execution versus the revenue generated - is something that many companies fail to do, not least because it can be difficult to identify all the costs involved, or indeed what activity has specifically generated what revenue. In some cases this may need input from Watson, in some Einstein will be appropriate, and in certain cases native Salesforce with the formula field will do the trick:

But if you don’t know how much each one of these things is worth, you don’t know whether you should bet on it, how big of an opportunity it is.

My take

The pairing of Einstein and Watson looks like it could give businesses something approaching the best of both worlds in the former’s ability to look from inside the business outwards, and the latter’s ability focus down the outside world to something pertinent and relevant to an individual business.