Author Archives: Daragh O Brien

About Daragh O Brien

Daragh O Brien is the Managing Director of Castlebridge Associates. This site has been one of his side projects for a decade. It needs some love and attention...

A process problem with the trains…

Information can take many forms. Ultimately, it is the ‘message’ that is communicated between two people (or systems or processes).

This story from today’s Irish Examiner shows the importance of checking that the message being communicated is accurate and timely. It’s a trainwreck because it relates to trains, there was a problem and it will cost the Irish rail operator €10,000 to compensate people for a miscommunication and an error in the presentation of information.

The summary of the article is that

  1. Over 300 people were waiting for the train from one of Ireland’s flagship railway stations.
  2. A visually impaired passenger was helped by the Station Controller on to the train
  3. The signal for the train to depart was given, while 300+ people stood on the platform.
  4. The people at the station had to take a different train (leading no doubt to over crowding), which made an unscheduled stop to link up with a special shuttle train transfer that brought them to their final destination.

Inaccurate information was given to the train driver and guard (or perhaps they interpreted information incorrectly) who followed the procedure when that signal was given – they started the train up and left the platform (and the 300 people) behind it.
How might this have been avoided? As ever in these cases an investigation is underway.

The focus of quality is the Customer

I got an email into my work inbox today enticing me to attend a Gartner CRM summit in London at which Don Peppers (he of Peppers & Rogers, 1-to-1 marketing fame etc.) would be keynoting.

I won’t be going.

The email was sent to my email address from a site that espouses CRM best practices.

The email salutation was “Dear Test”. Click on the image below to see the full screenshot.

peppers and rogers email error

Yes, it is easy to make mistakes with direct mail and direct email (the IAIDQ has had its moments in the past, but we work hard to understand root causes and prevent errors). However, where information is missing from a profile a valid ‘default’ should be selected (“Dear CRM Practitioner” or something”) . Referring to your customers as “tests” means you’ve failed a test yourself.

As an aside, a substantial root cause of the 70%+ failure rate of CRM implementations has been a failure to tackle the issues of poor quality information.

Of course, this isn’t the only email I’ve received recently where an Information Quality issue makes me discount the value of the email. Gartner continually send emails to my work account addressed to Ms. Daragh O’Brien. This is not correct.

Of course, when I click on the link in the email to update my profile with Gartner, I should be able to change this and fix what appears to be an error caused by some data cleansing tool. Unfortunately not, I can only change my marketing suppressions.

Another organisation in the UK continually changes the address they have for me to an address in a building the company I work for isn’t based in anymore and which I never worked in or gave as a contact address. This is obviously master data they’ve bought from somewhere but try as I might they don’t seem able to correct it.

Errors will creep into processes and into data. The best approach is to make sure you don’t have a process that creates poor quality information (inserting “Test” in the salutation rather than a name) or that you provide opportunities for people to correct your information (such as correcting name prefixes that are wrong).

Home Removals – literally.

Fox News has this story from the AP Newswire today.

It seems a Russian woman returned from a visit to the country to find that her city home was gone, demolished by mistake by over-eager builders who were supposed to be tearing down a different building.

Ooops.

Of course, errors in demolition can only happen in Mother Russia. Surely.

Apparently not. Looking back to August of last year, it seems that in the US, there is a bit of a muddle in New Orleans about what buildings damaged in Hurricane Katrina should be on the demolition list. According to Associated Press/MSNBC,

“Homes that were only damaged have wound up on a list of 1,700 condemned properties. Some houses on the list have been gutted for rebuilding or are in move-in condition”.

According to a spokesman for the Army Corps of Engineers, there have been cases of ‘do not demolish’ notices issued for buildings that they’ve already bulldozed because they were on the ‘demolish’ list.

The American Bar Association also picked up on this issue in August last year (which means the lawyers are circling… always a sign of a trainwreck). And in a case cited in the Wall Street Journal (subscription required) a homeowner was told by city employees that their home wasn’t on the demolition list and then, effectively, went home to find it demolished after she had spent money clearing the property for renovation and rebuilding.

My grandfather was a master plasterer and carpenter. One of the most important rules of thumb he taught me as a kid was “measure twice, cut once” to avoid waste and rework. Does that rule apply to demolitions as well?

Customer data boo-boos in Carphone Warehouse & Talk Talk

For a change, we didn’t find this one on The Register (oh, hang on , here it is on el Reg as well…). However, it would seem that UK communications retail Carphone Warehouse and its telco subsidiary Talk Talk have been given a stern reprimand from the UK’s Information Commissioner for problems with the quality of their customer information which resulted in breaches of the Data Protection Act. In addition incorrect information was sent to credit referencing agencies and debt collection agencies.

The full details can be found here: http://www.computing.co.uk/computing/news/2207387/carphone-warehouse-breaches

This echoes similar issues in the Irish Republic a few years back where Talk Talk’s Irish operation was reprimanded by both the Irish Data Protection Commissioner  for their information management practices. In that instance they were ordered to refrain from any direct marketing until they had sorted the problems out.

Dream host, Billing Nightmare

Courtesy (yet again) of The Register comes this case of poor Information Quality. It seems that US web hosting company DreamHost accidentally overbilled its customers for services due to what has been described as a “fat finger error”.

Full details of the good intentions that paved the path to this Information Quality Hell can be found on the company’s blog – they are refreshingly honest, if perhaps misreading the seriousness of tone that these type of issues require. Also some questions appear to be still unanswered (like how did some customers get billed twice for future dates). The ‘official story’ can be found on their Status site. On both sites the comments illustrate the impact on their customers.

Why is this an IQ Trainwreck? Well, by the company’s own admission, nearly every one of their customers has been overbilled. Many of these customers may have incurred additional bank or credit card charges if they have exceed overdraft or credit limit thresholds – which will probably have to be refunded by Dreamhost.

The root cause – a fat finger that created parameters for manual rebilling checks that were in the future… 2008 was the year keyed in, not 2007. And their billing software did not contain a business rule to either prevent or validate any attempt to bill for a future date.
Dreamhost fail to identify the need for a proof reading check to ensure that data going into a process (such as dates) fall within reasonable bounds for the process (choosing of course to blame the software). Of course, many of the 415 commenters on their blog have picked up on this simple step that could have avoided this IQ Trainwreck.

However, Dreamhost’s handling of the situation reveals another ‘cultural’ issue that means that these types of problems will recur. Their focus has not been on the customer – while some may appreciate the jokey tone of their blog post explanation, many of the commenters on their blog have condemned their ‘jokey’ if honest posting about the issue (it is perfectly OK for IQ Trainwrecks to joke about these things – we want people to laugh and then think – ‘oh, that could happen to me’).  As one commenter put it

“Hey, sorry your rent bounced, but here’s a picture of Homer Simpson and some lulzy hipster prose. Joking around might not be the best technique when you are messing with people’s money”.

Finally, it is an IQ Trainwreck because Dreamhosts competitors have jumped on the opportunity to steal business from them. One of their competitors has created a discount code for people switching hosting to them from Dreamhost which gives them savings on their hosting costs (with no guarantee they won’t be as clumsy with their billing I suspect).

This will be a costly one to put right.

I’m wrong to the finish, coz I eats my spinach…

Every one knows that spinach is good for you. However it isn’t as good for you as initially thought.  Vincent McBurney over at ITToolbox.com wrote a post last year about the ‘spinach iron content controversy’ and its effect on people’s thinking about the health benefits of the leafy vegetable.

This qualifies as an IQ Trainwreck for many reasons, but the main one for me is that it means most of my childhood Saturday mornings were based on an IQ error… Popeye could never have gotten that strong eating spinach.

More importantly it teaches us some valuable lessons about the impact of poor information quality. Firstly, it can happen very easily (in this case a decimal point seems to have been in the wrong place due to a miscalculation).

Secondly, there may be a more complex ‘system’ at work that might affect the real business impact of the information (1000 orders pending might sound bad, but is it bad if they are in dispatch and the order will complete once the goods are delivered?).

Finally, the ‘conventional wisdom’ that pervades through an organisation can often be difficult to overcome, even when it is based on poor quality information and an incomplete understanding of the system (or information chain).  Even today there are many people who think that spinach is a good source of iron in the diet, even though the system is complex (only 2%-5% of iron in spinach is absorbed) and the iron content is lower than many people think (90% lower to be precise).

Read Vincent’s post for more details.

Blog Carnival of Data Quality

In a break from our usual editorial remit here at IQ Trainwrecks.com, we are pleased to host the second edition of the “Blog Carnival of Data Quality” initiated by Vincent McBurney back in November.

Given the volume of posts we had submitted, and the range of relevance to the theme that Vincent had suggested for this month’s Blog Carnival, some pruning has been done on the submissions that came in. In Information Quality terms, that’s kind of like inspecting defects (non-relevant posts) out of the process.

Honourable mentions

Some posts were a bit more on target, but didn’t speak to an Information/Data Quality theme sufficiently clearly. That said, we can learn a bit about only storing the information we need and archiving stuff from Richard Lee and his PDF Black Holes. Likewise, John W. Furst’s advice about disaster recovery, while not strictly relevant to the Data Quality agenda, should have some relevance to all (availability/accessibility are often cited as quality critieria for information).

Gold Medal Blogger

Beth Breidenbach was quite busy over the past month.

She created an agreggator for Information Quality blogs. She blogged about the new education programmes in Information Quality. She wrote about Quality in unstructured data and a tonne of other things. Taken together, Beth’s posts show the coming challenges in IQ/DQ (unstructured/semi-structured information), the beginnings of a professional discipline with strong academic foundations, and the importance to organisations of the quality of their information. All of which are important themes going into 2008 and beyond.

The Others…

Steven Sarsfield shared with us his “Winners and Losers in Information Quality“, where he highlights the importance of good quality information to the ‘goodwill’ a company has with their customers. Also, who’d have thought that improving the quality of your information might help save the planet?

I wrote a post over on my blog, the DoBlog, addressing what would make me happy (from an Information Quality perspective) in 2008, and looking back over 2007 at what went well in the Information Quality world. Hopefully it will provoke some thought and/or debate on the direction of the profession.

Finally, Vincent shared his look back at data integration software in 2007 – a year of performance improvements and consolidation for the big vendors and a continuation of the open source ETL gold rush.

That concludes this edition. Submit your blog article to the next edition of Carnival of Data Quality using the Carnival Submission Form.</a>

Past posts and future hosts can be found on the Blog carnival index page.

A Cat-astrophe

December is the month when the “wacky news” stories from around the world get an airing…

 I found this one via Yahoo this morning… Cat gets credit card.

An Australian woman decided to test the indentity screening processes that her bank uses for credit card applications. So she applied for a card in the name of her cat, who is 2 years old.

The bank asked for ID documents, but hadn’t received them before they issued the card (the cat was getting preened for the passport photographs – we can only assume). Furthermore the cat’s owner wasn’t notified that a second card had been issued on her account.

AUS$4,200 of a credit limit is available to the cat.

Trainwreck or Bus disaster?

From the “This is london” website there is this story about the dangers of relying on satnav and the unexpected outcomes that arise from decisions taken on incomplete or inaccurate information.

Summary… 50 people take a shopping trip from Gloucester to Lille (France). Driver of coach relies on SatNav. He selects “Lille” from the menu. He neglects to check if it is the right Lille as there are two within 100 miles of each other – one in France and one in Belgium. Coach tour winds up miles off course in the wrong country.

The passengers notice this as they are travelling, particularly when they see signs for Eindhoven (not in France) and (possibly) “Welcome to Belgium”, but when they brought this to the driver’s attention they were ignored. The end result was they wound up in the wrong country and, by the time they got to the right Lille (France) there were only 2 hours of shopping time left.

The shoppers stayed overnight in Lille and the next day the driver turned off his SatNav so as not to make the same mistake again. Unfortunately that resulted in him getting stuck under a low bridge and having to go back and find an alternative route, adding an hour to the journey.

What is the moral of this story?

  1. When using technology in a process, pay attention to signposts along the way that could tell you you are going the wrong way.
  2. Listen to your customer… sometimes they see signs of error you don’t
  3. Don’t blame technology for errors originating with people or incomplete information. Rather than throw out technology, first look at the process end to end and see where the problem actually originated. It might mean you can use the existing technology more effectively and avoid future delays and problems.

Is it an IQ Trainwreck? 50 people affected, poor quality information involved, a double whammy of issues…. if not a trainwreck then perhaps a Bus Disaster?

So, they’ve got guns and are trained to kill…

let’s screw with their pay…

From the erstwhile The Register comes this story about on-going information quality problems in the British Forces pay and personnel system. There have been complaints of pay being withheld for months.

The MOD blames the data input monkeys  staff and insists that the system is working fine.

“Input errors based on a degree of unfamiliarity with the new scheme have resulted in a small number of pay inaccuracies,” according to the MoD.

“Thorough investigation of these errors has shown that the JPA system is working extremely well… JPA… requires accurate and timely input from… HR administrators.”

Additional training measures are, of course, being provided to staff and the MoD is keen to point out the long term benefits of the system in terms of reduced manpower needs in HR and fewer systems to maintain.

Of course, the complexity of the payroll system should not be underestimated, particularly if there are staff at similar grade with differing pay structures. Add to that the requirement for rock solid security (given the sensitivity of the information) and the system requirements become even more complex.

However, basic validation and verification of information (perhaps a reconciliation between the new system and the old system at data migration) might have mitigated this problem.

Why is this an IQ Trainwreck? Well, they’ve pissed off members of some of the most elite fighting units in the world… not something I’d do.