The role of technology in a data-driven strategy
What can we learn from Hyatt?
Now, the Hyatt example is just one of many ways in which data can be used to drive better business outcomes. The key point here is that this has nothing to do with the:
- Type of data that they chose to use
- Technology platforms on which they have housed this data
Instead, this is a story about how the Hyatt has been able to take data and support a key business philosophy and a key business strategy, which is to treat each of their customers as individuals and to look for ways to create unique experiences, to delight those customers as they stay in their hotels.
As they deliver on this promise to create unique, memorable experiences, they’re then deepening the connections with their customers and fulfilling their strategic objective of being the most loved hospitality brand in the world.
So where does technology play a role in all of this?
Well, here’s the high tech version. Let’s imagine now that somebody at the Hyatt has hooked up to Facebook’s application programming interfaces (APIs), and they are automatically extracting the text and the images from my posts – plus thousands or millions of other posts globally.
They’re then parsing that data through something like an NLP or natural language processing algorithm, which is able to pick out things like the entities – so “Hyatt”, me as a customer, my name, products that they might be offering. It can also pick up potentially a sentiment analysis as well, which is going to then tell you, is this customer happy? Is this customer sad? They could also potentially look at the image data itself.
Now, Facebook hides the EXIF data. As far as I’m aware, that tells you the GPS locations of the person. That’s in the metadata. However, they could easily run it through an algorithm that can compare against their own property image data and work out whether or not this is a match.
So this is a high tech version of finding out what it is and flagging this as a potential piece of information that could be relevant to their strategy. They could then import that into their corporate systems, their ERP systems or their CRM systems, use that to alert the local duty manager, who then makes the changes and updates the systems of record that are used by the kitchen and catering teams. So that’s a High-Tech version. That’s costly. That requires thought, time and effort.
The low tech version of the same thing
If you really wanted to do this manually, you can hire an intern here in Singapore for less than $1,000 a month. That intern can join the Facebook group that I was a member of. Now there are probably about 5 or 10 posts per day that are coming up on there. You can also have them looking at other posts from regular guests as well.
That intern spots a meal that looks a lot like the meal I had. Learns that I’m unhappy, sees my name and proceeds to check the internal CRM that states I’m in room 886.
Easy peasy. Again, call the kitchen. Tell them to do a better job. You don’t even need a system for that. So there are two vastly different approaches. And they have two vastly different cost structures associated with them.
The key thing to learn here is that, yes, you are supporting a key business strategy. That key business strategy has a tangible, valuable outcome for the business. There is a cost associated with solving the problem. In order for you to work out whether or not you should pursue a data-driven strategy and whether or not you should go high tech or low tech, you need to work out:
- What’s the value to the business?
- What’s the cost to the business?
If the cost is less than the value, go right ahead.
There is a tendency for some in data teams to overengineer solutions to problems. It’s often driven by a desire to play with the fun toys or to get some new skills on your CV. And that’s admirable. It’s definitely something to shoot for. But just make sure that you’re spending the budget of your organisation wisely. Because let’s face it, it’s about delivering business value. Don’t go out shopping for some fancy flashy AI or machine learning algorithms, if an intern, a phone line and an Internet connection will do.