Complexity’s Competitive Edge

How does a global company such as Intercontinental Hotel Group (IHG) get its individual hotel operators to embrace analytic decision making? How does it decide how and when to deploy its search-engine bid-management platform? How does it keep competitors from borrowing the techniques it builds?

MIT Sloan Management Review raised these questions with four leaders across IHG’s Global Sales and Marketing team: Larry Seligman, vice president of advanced consumer analytics; Jim Sprigg, director for database marketing and analytics; Angela Galeziowski, vice president of worldwide sales, strategic insights and planning; and Dev Koushik, vice president of global revenue optimization.

It comes down to managing complexity, the IHG team noted. If you can effectively address complexity in modern marketing, you gain a competitive advantage that can take a lot of time for competitors to replicate.

The interview was conducted by Sam Ransbotham, an associate professor of information systems at the Carroll School of Management at Boston College and the MIT SMR guest editor for the Data and Analytics Big Idea Initiative. Answers by individual IHG team members have been bundled together.

Hotels and airlines were early leaders in using analytics in their operations. What is IHG doing today that you think the rest of the world would find different and interesting and novel?

We have done a tremendous amount of work in trying to focus our efforts towards how to improve pricing performance. This includes things like discounts and anchoring-off our best available rate. We work with a lot of regional revenue management teams and the hotels at the unit level, to educate them on what right pricing strategy and pricing structure should be in place, and to give individual hotels ways to set prices in an automated fashion.

We developed a capability called Price Optimization, which revolves around some advanced analytics. We can embed the current business practices and account both for our demand for cash for the next 365 days and for the price sensitivity of the retail demand to different price points. This capability also accounts for what comparative rates look like. All of these things are done in an automated fashion and come up with an optimal rate at the property level for every day for the next 365 days.

I’m guessing that your property managers have varying degrees of analytical savviness. Is there difficulty getting them to buy into those predictions? You’re certainly balancing an intuition versus data approach.

Absolutely. The way we approach this is very practical in a sense. We have to put together a prototype they can easily interact with and understand, and we package it up in a format that allows them to focus on the key things, rather than exposing everything under the hood.

Not every revenue manager nor every hotel manager is initially willing to use those things, so we have to put in a lot of effort to make sure that they understand the exact business process. We first rolled this out in 2009 and we have been very successful in building it out to almost 4,000 properties today.

Deciding where you’re going to allocate your marketing dollars usually occurs at the corporate level. Does it also influence the pricing model?

Yes. There is a reason why we do revenue management at the hotel level. We want our hotels to understand the true value of retail demand. But the marketing dollars we put in for some regions would be automatically reflected in the pace of the bookings that come through. Hotels would pick up on that immediately and then have to consider, “How can I price it better?” and “How can I manage my revenue so that I can maximize our share of the marketplace?”

A lot of our marketing is around clusters of properties, especially around cities and geographies that are destinations, and of course we use analytics for the decisions we make there. If we have a market where it looks like it’s going to be sold out, then we’re not going to bid as aggressively in our search engine marketing for key words around that market. On the flip side, if there’s a market in need, we can use those search engine buys. So there’s a positive feedback loop that we’re working to develop. We haven’t gotten there completely yet, but that’s definitely something that’s in the forefront of our thinking.

Why isn’t that developed yet? What makes this difficult?

It’s the complexity of it. It’s getting the right context. Within search-engine marketing, you don’t necessarily have the clarity of information that you get from revenue management.

Let’s say Houston is really hot right now and oversold for the next week, but the week after, business is significantly slower. If you, as a consumer, are searching hotels in Houston, I, as a marketer, don’t necessarily understand whether you’re looking for next week or the week after. We’re now thinking about how we can incorporate that data into where the need state is more apparent, around, say, meta-search on Trip Advisor and the Google product listing version of that called Google Hotel Ads. With those, you put in dates and locations, and we have more specifics that we can use to figure out the value.

Optimizing search-engine marketing does sound like a challenge.

Yes, but we’ll say more broadly that another challenge is how to operationalize this information. For example, with attribution, we’ve launched a digital attribution platform using a third-party vendor. That attribution system is integrated with our search-engine bid-management platform so that our bid management is optimizing on the attributed revenue for every one of our millions of key words that we’re buying. That’s opposed to the last-click revenue, which is what they were doing before. That whole operationalization is a key challenge and something that we think we’ve really moved the needle on. I don’t think there are many hotel companies doing that kind of sophisticated, high-volume algorithmic automation around search-engine marketing.

A lot of what is done in analytics is around discovering some novel way of doing things, such as marketing. There’s a lot of effort that goes into doing that, but other people can see your work. How much do you worry about that? Or is the complexity enough to keep people from copying it?

That’s a really good question. In our work targeting customers, we use predictive modeling and a lot of data and profiling and optimization — for instance, using marketing to focus on our high-value customers.

Our success has a lot to do with the way we’ve produced our marketing. It’s highly individualized in terms of routing different types of offers and promotions to different customers. Basically, we’re trying to optimize actual lifetime value over a period of time, measuring, for each individual, what are the incremental hotel stays or the incremental brand experiences — optimizing several things at once. And that’s something that seems to be hard for some companies to copy.

The science behind marketing to known customers has become a bit of a lost art. And it’s a lost art because of big data and the complexities of multi-channel marketing. A lot of companies try to use methods that were standard in the past, but they’ve had a tough time evolving those methods to deal with the complexities that we have, where we have many stakeholders trying to drive many types of behavior among the same customers with many brands across multiple channels.

By focusing on trying to crack the nut on the complexity of marketing in the modern environment, that’s where we think we haven’t seen our competitors start to copy our approach. They can see what we’re doing because they can look at our promotions. But addressing complexity, if you can address complexity in modern marketing, gives companies a competitive advantage that can take time for competitors to replicate.

Companies who are successful in these things also need an organization-wide focus on analytics, or at least a commitment to it, too.

Yes. Having the ability to replicate something and actually doing it are two different things. In order to actually do it, you’ve got to have leadership that believes in it, is willing to put resources against it, and is willing to change current processes. They have to be willing to replace something they know, with a new and what they could see as a potentially risky approach. That demands an organizational culture to build those capabilities and have them embedded, and to transfer existing processes over to them.

Also, even when you replicate, it’s not going to be exactly the same, because there’s a lot to analytics that is art and not science. You’re going to have people taking different perspectives on how to approach a given problem.

You mention the art of analytics, and having a culture that’s ready for big changes.

Those are both really big deals. The biggest challenge is not so much coming up with the models and coming up with the data. It’s engaging the decision makers and working within the culture.

Part of the success of doing that is recognizing that this “intuition versus data” idea is a false dichotomy. Great analytics teams love intuitive thinkers who love data, because it’s that intuition and that human spark which brings insight and innovation. And so, when we work with a decision maker and we identify the questions we’re trying to answer, it sparks a whole new conversation. Just because someone has good gut instincts doesn’t mean that they’re your enemy. They’re really your best friend.