Data'll
BARBARA Clift can predict which of your customers will not honour their reservations. She claims that her predictions are 95% accurate and that she can do it so well that airlines want to buy the secret from her. Now her company, IBM, wants to sell the secret to you.
"We can predict with incredible accuracy the lifetime of a customer's relationship with you," says Clift, segment manager for IBM's travel-related division. "So you may choose to treat a customer differently because they are going to be of high value."
"How do you do it?" is the question Clift really enjoys. Using maths, she says. Maths, plus IBM's vast experience of mining customer data, experience gathered during 40 years of building database systems for retail chains, airlines and a range of other consumer businesses. "All these tools can apply to other databases elsewhere," she says. "If hotels could do this, it would enable them to optimise their rooms with more accuracy."
What Clift is talking about is how to refine the mountains of data that hotel staff tap into reception desk and call-centre computers each day, and turn it into something which hoteliers can use to cut their costs.
This isn't just yield management or mailing lists. The data is stored in a form of database called a warehouse, and getting useful information out is known as data mining, in today's marketing. If all of an establishment's databases are integrated, it makes it easier to "mine" the data for the useful information.
The distributors of property management and central reservations systems may claim to already use that data efficiently, but data warehousing specialists think that data contains more secrets about your customers - secrets you can use to make your life easier.
Integration
Why might you want to do this? Because it will allow you to exploit some of the features that are now showing up in front-office systems for individual properties. For example, if you've been using separate systems for your front office and restaurant, the databases are likely to be separate. Integrating them could let data warehousing experts find out which of your guests spends how much on what. Knowing that will let you customise your marketing and promotions around what really turns those customers on.
One supplier, Performance Technologies, has linked with Sharp's EPoS system to enable billing of a guest's restaurant spend automatically, which in turn will allow data to be collected on spending habits.
At the cutting edge of data mining are people such as Clift at IBM, who believe that they can extract much more personal customer information and that it will tell you about your customers' plans, likes and dislikes. In 10 years this may be passé but, for the majority of operators, it is futuristic and the present is somewhat different.
"A lot of property management systems don't let you interrogate the data," says InnSite's sales and marketing director, Phil Davidson. "What we're working on is overlaying our package, Corporate Sales and Marketing [CSM], on a property management system [PMS]." InnSite has sold CSM to big hotel chains for years but the complexity of the package has always pushed its price beyond the reach of smaller hotels.
CSM will produce six standard reports showing average room rates, average occupancy rates and average stay rates, but can extend beyond this to something around which hoteliers can build a real marketing campaign - based on what that hotelier's customers really want. It will not be available for individual properties until early 1999 but the benefits of warehousing are beginning to cascade down from the largest chains.
Marketing and purchasing consortium MinOtel, for example, consolidated the six databases it was using for its 200-plus UK hotel clients into a single database earlier this year - an enormous task carried out by specialist Infocentre, which wrote the new database for the purpose.
This may have affected the consortium's marketing activities this year, acknowledges director and general manager Sally Boughton, but it was necessary to grow. "But," she says, "we're now able to pull together sectors of our members that will suit promotions aimed at sectors of the marketplace, and so we're able to promote them more individually, as well as on a group basis."
Being more targeted means that MinOtel can more easily reduce the money it spends on funding promotions to customers outside of a particular sector and who are unlikely to react. So warehousing its data cuts costs.
There have been other advantages too. "When we looked at the fields of information we had," Boughton says, "we saw there was loads that we didn't actually need, so it's given us an opportunity to clear that out." That means that when you ask staff to enter information, or even when you specify what you want of a PMS, you're likely to have a much better idea of which pieces of information will actually bring you a return.
Being able to spot the real returns you get from different bits of information is at the heart of warehousing data and mining, Davidson says. But the ability to do it may come at a high cost. "You'll be worrying about the cost of the IT if you make changes to systems that work," he says.
However, making those changes could save you some of the real costs incurred by problems with the data you already collect. It gets complex here but, by high costs, Davidson means the cost of getting the information you already have into a form that will allow you accurately to apply the filters, queries and maths to analyse it properly.
Normalisation
That process - it's called "normalisation" in the geek-speak of the database industry - is likely to come with a big price tag, because the databases underneath the hi-tech front end of most PMSs are a mess. They're strewn with the kind of mistakes that happen when busy receptionists miskey, or get called away while correcting an earlier error.
Real world PMS/CRS databases have address details sitting in customer name fields, postcodes sitting in telephone fields, and all the other things that make a nonsense of any effort to cost-effectively mail customers with details of new deals.
Those are the real, if hidden, costs of current hospitality databases Davidson thinks you can escape by cleaning up the data now.
Database vendors can do a lot of that for you automatically but checking that the automation worked correctly is necessary and expensive. "It's so expensive, you're better off starting from scratch," says Davidson. In other words, scrapping the data you already have - other than a basic customer database.
Well, he would say that, wouldn't he? you might think. After all, he's got a product to sell and doesn't want hassles. And you'd be right - but every other database vendor or consultant agrees. They're stunned at the messes they find when they start looking at the databases hoteliers are creating with their day-to-day use of PMS/CRS software.
Barbara Clift agrees and so do the other vendors - they all suck in a deep breath when you ask them whether it is easy to transfer data from hoteliers' existing databases.
If you take Davidson's recommendation that you should start from scratch - and the consensus is that you should - you are going to have to factor in the cost of starting without much data. But the long-term returns could mean that it is a price worth paying. n
