23 January 2011

Bad Search and Not Fair Search. Quality and Reference Point Matters

There has been a lot of talk about consumer fair search and the value of the travel agent in the mix.

The rhetoric is loud and raucous on both sides. Let me make one thing completely clear - the Professor is WHOLE HEARTED in his support of the agency community. I firmly believe that the neutral channel is viable - indeed essential - in today's environment. However where I differ from the so called Open Allies Group is that I believe that the tools provided into the market are bad - IE Legacy GDS solutions are just inadequate and fail to provide the consumer choice. The old story of GARBAGE IN GARBAGE OUT applies no more clearly than in Travel Agent Search Tools from the legacy GDSs.

The value an agent is supposed to bring to the consumer is that he can interpret the possible results and help the consumer to make the best choice possible. But what if the tools he uses are not that good? Is the consumer then best served?

To illustrate my point here are 3 different results sets from Delta.com, Orbitz and LUTE Technologies Agent tool YuFares powered by Everbread.

The query is the same and should therefore represent the same basic result no matter what. My designated query here was as follows:

Seattle to London - simple roundtrip
Outbound Feb 6th
Inbound March 18th
Price led search

However the results shown here are constrained by certain parameters.

Let's look at the airline direct result first. Delta.com - I use this site a lot because I fly DL a lot. However I know the constraints on the site. For example in this display I want see what I can get. There are 20 results. The shortest elapsed time in the 20 results that Delta.com returned to me was 23 hour and 45 mins. The longest time was 35 hours 17 mins with a 22 hour and 10 min layover. The price range of the results was from $731.70 to $758.70

NOTE I believe that Delta uses the Travelport search system. Travelport also owns Galileo and Worldspan and is currently embroiled in a legal fight with American Airlines.

Next I looked at Orbitz. This is the master Orbitz site in the USA. Same search parameters. After getting the master display - I used the Delta only results - there are considerably more possibilities that I could have chosen. According to Orbitz it gave me 24 Delta specific choices (including code shares and interline with partner results -AF and KL). The time line ranged from 12 hours 30 to 15 hours 20 and the price ranged from $756 to $1934.

Note: I believe that Orbitz uses the ITA Software engine. ITA Software is involved in a takeover by Google. Orbitz is embroiled in a fight with AA and is 48% owned by Travelport.

So finally I looked at one of our clients LUTE Technologies whose product YuFares uses the Everbread engine for B2B and B2C customers. The version I used was open and used the B2C standard interface. I will ignore that B2B Agent version as this found fares cheaper than the previous sites with some complexity. While the complexity of the trip was within the boundaries of what the other sites found and would be normal for an agent to offer a customer - I thought it would be fairer to use the untuned version as the results were within acceptable boundaries.

The system returned 111 results of which it displayed 15. The timeline range was from 12 hours 30 to 16 hours. The price range was from $769 to $887.

Note: I investigated the results for each of the hosts and I can see that Worldspan which powers Orbitz actually constructs a lower fare than the LUTE results. For the LUTE results - the engine can choose from any GDS - so I selected a different GDS to highlight the difference. In this case I used the Amadeus GDS for availability options. When switching the engine to the different host - LUTE and Orbitz match. I indeed checked this on ITA's own mobile app and it too confirmed the results.

The clear conclusion one can draw is that for a simple search the results can vary rapidly. The fact that Delta seems to use an inferior tool for its search results is something that Delta will have to address as they go through the process of upgrading its site. For Orbitz (ITA) and the LUTE Technologies (Everbread) engines - it can be seen that the consumer benefits from a better tool.

Based on the results: -

Who had the cheapest fare?
The airline

Who had the optimized search?
Orbitz and LUTE tie with a nudge to Orbitz for the cheaper fare using Worldspan (See note above on host systems).

Who had the best elapsed flight timea?
LUTE and Orbitz tie

Who returned the fastest result?
Lute by a hair. Orbitz near enough as a tie. Airline result was much slower.

Clearly if the legacy GDS is the system powering these results then one has to ask - what is fair search?

If you evaluate the consumer position would the consumer actually make a decision that has him sitting at an airport for nearly 24 hours to save a whole $23.70. That would easily get consumed in a few trips to the airport junk food court.

The possible results from availability differences are clear. If you use the GDS tools then you are likely to see different results as each GDS is constrained in what it can provide as a set of results. It is not that the answers are not there - it is that the GDS cannot display all the spectrum of possible answers. Newer technologies such as ITA and the YuFares powered results clearly demonstrate this. The brute force solution of lowest fare no matter what shows that there is a subtle case for displaying results based on business rules. The legacy GDSs cannot do that.

I hope that this adds to the debate and puts the issues of search in context. In my view search is not a simple thing and by now anyone who has taken more than 5 minutes to assess the situation must understand that Available Airfare search is a pretty hard and complex thing. From a consumer perspective with all this complexity - Google making an entrance can be seen as positive.

For the consumer the issue of search is cannot be just about price. But making an informed decision in a transparent manner is not trivial. The chorus of those who say that the legacy GDS is the be all and end all needs to get real about the issue. The issue of Direct Connect should be about quality of result and service to the consumer. Sadly that seems to have been forgotten. All these organizations claiming that legacy GDS powered results are essential to the framework of freedom of choice is pure hogwash. I don't think there is any 1950s/60s technology that should - or should I say COULD - ever make that claim. GDSs can do somethings really well. Search just happens not to be one of them.

So what do you think?


Anonymous said...

Great post, Professor...

Just a side note: it seems that delta.com miscalculates the airport taxes - the actual air fare is the same.

Maybe this is another incentive to push the customer to airline.com: book with us and we will pick up the tab on the wrong tax calculation...


Professor Sabena said...

I guess then where you buy matters. This confirms one of my points that the calculation of taxes can be a less than precise service. It seems if my Anon friend is correct that Delta is not charging all the taxes. So the DL result will be less by the differential with Orbitz and the LUTE YuFares result.

If an agency did that - they would be subject to an ADM - Agent Debit Memo. I think DL cannot charge itself for its own mistakes - but I bet there is someone out there who should try and determine if the US government or whomever the taxation authority is - may be getting cheated out of tax dollars.