Date: 18th May 2012
I’m just back from an enjoyable time at the UK Computer Measurement Group (UKCMG) annual conference.
Aside from the chance to hear cool presentations from the leading lights of the international CMG community, and to swap war stories with others who spend their time deep in Capacity Management, modelling, system/website load testing and all the associated challenges, I experienced the most engaged and audience for a presentation ever ! There were so many questions for Innes, they were reluctant to call a halt!
Innes Murray, IT Director of Inghams Travel, was on the podium, with yours truly adding 5 minutes of detail in the middle. He opened the eyes of those present with his tips on measuring and managing the issue of Unbuyable Holidays (the Caching Problem he called it) to maintain best customer experience and purchase rates.
Innes went into specific depth on how measuring live usage through dynamic user journeys had brought focus across both the business and IT teams with a shared interest in customers experience. Of course this experience is directly dependent upon the shifting sand of the dynamic cache that is itself based on 3rd-party data (flight availability in this case) that itself is dynamic and only available from suppliers as a cached record.
In simple terms the challenge is this – reducing the bad customer experience that results from a product that is find-able on the website from cached product data, but is then not actually buy able at checkout when a realtime availability check is made.
The data caching solution itself was arrived upon initially by the travel industry as a way to improve user experience by removing the need to run many realtime queries that would slow down the user search.
In Ingham’s Travel case, the nightly list (cache) of available holidays comes to over 7M records based on combinations of destination chalets, resorts, starting airports, number of days etc.
While this cached list makes search faster and more responsive for the user the data is all “old” even if only by a few hours. This means that user searches during the day can sometimes produce results where a user can find a holiday, but in the later pages of booking and adding their details are then told that it is not in fact available.
This can for example when the necessary specific schedule flights for that package are no longer buyable from the relevant air-line. Air lines often release seats in blocks, not the whole aircraft. For a specific flight there may be no seats at price £X on one day but some new ones may become available the next, only to sell out again within a couple of days and once more be unavailable for a few days until the next ‘block’ is released, perhaps at a different price point. Users do not know or care about this, they only feel frustrated that they could not book the holiday they thought they had found, and then are confused when, having been told their combination of requirements was sold out, or otherwise unavailable, see the same thing offered again when they visit the site at a later date.
These customer journey ‘dead ends’ are very damaging to sales.
Some of these dead-ends are also due to technical issues within the eCommerce website itself, as well as within the inventory management Cache as above.
Innes explained that through the use of realistic dynamic user journeys, that choose a different package each for each sample every 5 minutes 24/7 he’s able to measure the ‘dead ends’, and address the technical issues in his website and inventory management caching systems, to reduce them.
Despite the fact that Innes had the ‘graveyard slot’ after lunch, the presentation hit a nerve even for the predominently non-travel audience, who experienced many of the same core issues with their own inventory and subsequent planning, prioritisation and capacity management.
It was also interesting to swap selenium performance testing consultancy stories – and other interesting speaker slots on Monday were:
- the Keynote and later session from Glenn Anderson, from the USA, on performance measurements of web systems in the Cloud – something that in the team here we are doing on a more and more frequent basis;
- Paul Offord of Advance7 on building an effective cross-technology Rapid Problem Solving team – something where the right measurement tools providing the right evidence are vital to the speed of solution.
- The Capacitas guys as always did some useful sessions – Hi Andy !
Contact us if you would like to find out more about how our load testing, inventory management and support for capacity planning.