Data is in the limelight—again. Similar to the “big data”
phenomenon that gripped the travel industry a decade or so ago ("one of
the fastest growing and most talked about technology trends,” observed
Phocuswright analyst Bob Offutt back in 2012), today the wealth of artificial
intelligence (AI) applications has brought it back into focus.
But this time around it’s all about cleaning and
standardizing this abundance of data, as the travel industry shifts towards the use of AI agents. The
nirvana is for these agents to make decisions without human supervision in the
future; but they’ll only succeed if they are trusted. And for that they need
accurate data.
Experts also argue “cleaner” data is needed for more
automation as travel agencies keep pace with demand. A new report from BCG predicts leisure travel
alone will grow from a $5 trillion industry today to $15 trillion by 2040.
Stricter data compliance laws are behind the data cleansing drive too.
Time to scale
“We've definitely seen a shift this last year in the data
quality market. Now companies have tried to scale up their AI proofs of
concept, they've been hitting the inevitable brick walls you hit with poor
quality data,” said Robbie Jameson, CEO of data quality platform Tale of Data.
“Garbage in, garbage out: nothing new here, but people need
to experience it for themselves to really get a feel for the cost. So as a
result we're finding more people are coming to us not needing to be told what
data quality is, but wanting to find out how our augmented AI platform
addresses it.”
Subscribe to our newsletter below
The number of AI agents on offer is growing exponentially.
Switzerland’s Umbrella
Faces specializes in standardising data and traveler profiles, and
works with 600 travel agencies, across 70 countries. It has a database of 13
million traveler profiles—one of the most complex types of data sets.
“There are multiple companies that specialize in AI for
travel agencies,” said Helmut Pilz, senior vice president at Umbrella Faces. He
cited Acai
Travel, a PhocusWire Hot 25 Travel Startup for 2024 and this year’s recipient of the TravelTech Show’s Trailblazer Awards, as one
of the forerunners in this space. But many new marketplaces are cropping up alongside bridges between the old world and the new including Apaleo’s Agent Hub, PredictX’s Cogent, which is described as a
“workforce” of AI agents and Dataiera, an AI application layer between language models and legacy systems.
“They all need correct data. What we can make sure is, with
our software and workflows around it, that the data is clean in the sense it is
identified and sits where it should be, and that it’s complete, in terms of
mandatory information,” Pilz added.
Costly mistakes
With corporate travel, there’s no room for error. Even
simple elements like a missing phone number, or an expired passport, can lead
to operational disruptions, financial losses and reputational damage. If a
traveler’s stuck at an airport, or can’t check into their hotel, the system
falls apart.
In fact, underperforming AI programs and models built using
low-quality or inaccurate data cost companies up to 6% of annual revenue on
average according to a study by data platform Fivetran.
And some 70% of the top performers in a recent McKinsey survey said they have
experienced difficulties integrating data into AI models, ranging from issues
with data quality, defining processes for data governance, and having
sufficient training data.
Pilz also argues travel agencies today need better data for
internal automation to improve operational efficiencies, while agentic AI
offsets the shortfall of capable and experienced [human] agents.
“One of the key challenges travel managers face is
resourcing,” said Keesup Choe, CEO of PredictX, which won this year's Business Travel Innovation Faceoff at the
Business Travel Show Europe.
“While corporate travel has surged, many teams haven’t been
able to expand to meet this demand, amplifying the need for scalable,
autonomous solutions like Cogent that can address these gaps efficiently,” he
said.
Missed opportunities
Scott Wylie, chief technology officer at travel management
technology platform TripStax,
which offers a dedicated QC (Quality Control) module, also agreed that the
challenge is only a fraction of enterprise-critical data is currently being
surfaced in AI models.
He argues poor data quality can erode trust and lead to
missed opportunities, whether booking
data for reporting, or the use of AI for hyper personalization for an
individual based around a rich and accurate profile.
“We are seeing TMCs and corporates tackle challenges at an
enterprise level,” Wylie said. “Especially across regions such as China where
data delivery, accuracy and cleanliness need process improvements. The
objective has to be to move toward a near real-time data environment, which
will have to bypass traditional collection methods.”
Meanwhile he thinks the travel industry is still struggling
with data fragmentation, and reliability, because of aspects like NDC (New
Distribution Capability) which “paradoxically introduced new complexities by
fragmenting how data is managed and accessed.”
Forecasting the
future
Revenue management is another discipline where there’s
pressure to ensure systems are fed the most accurate data possible. Getting it
wrong can be a hotelier’s biggest fear, according to Mehdi Soua, chief
information officer at Louvre Hotels Group.
In a similar vein to Jameson’s “garbage in, garbage out”
analogy, Soua offered the French perspective—“merde in, merde out”—while speaking on stage at the Global
Revenue Forum in Paris recently.
“You could be using the smartest revenue management system,
but if the data isn't clean, if the information we feed it doesn't smell good,
only bad things will come out, so to speak,” he said.
“And speaking of data, this also brings us to the General
Data Protection Regulation. We have to be careful about the data we're going to
feed, but also related to confidentiality regarding company information. Today,
everyone puts things into ChatGPT for testing, but many people put company
documents into it, feeding models that can be shared with competitors or other
companies,” he said.
Meanwhile, in the tourism industry, there are extra
challenges because data is constantly “moving”, and is partly based on
personal, emotional and contextual experiences, said Claire Robinson, author
of a new whitepaper called “AI agents for tourism: can we trust them?”
It proposes four levers to build a trusted tourism AI, one of which includes investing in quality, reliable, structured and contextualized data.
“Can travelers really rely on AI-generated recommendations
for their trips? Misinformation, data inaccuracy and algorithmic bias
compromise the quality and reliability of the suggestions provided, exposing
users to an overload of unreliable options rather than genuine assistance in
their decision-making,” the report concluded.