Some 40 years ago, a bus engineer at the then Teesside Municipal Transport drove past the Tees Transporter Bridge, a marvel of engineering for which TMT had become responsible. "The bridge is out of true," he reported. A specialist team of consulting engineers turned up with what was, for the time, highly sophisticated equipment. They spent a day taking measurements, adjusting their equipment, making calculations. They concluded that the bridge was, indeed, very slightly out of alignment. How, they wondered, had the bus engineer spotted it, from a car, in a matter of seconds? He answered "let's assume the river's flat... the relative angles are wrong." That's what we've lost. We rely on ever more complex systems and we forget the basics.

The more I look at the complexities that financial institutions have had built into their systems, the more I realise that time and money is wasted and that, often, the customers' experience is worsened by bad decisions and bad systems. Often the bad decisions and bad systems are as a result of a failure to understand what harms are being protected against.

We can take an example from outside the financial sector to show why it is important to understand how failures occur.

An online discounter developed an excellent web presence. It's easy to navigate, product information was readily available and pricing was keen. Indeed, in markets where Amazon.Com has no or very limited presence, it developed market positioning that it was hoped would stand it in good stead in the future.

However, the big question that seemed to have been overlooked is what is the primary reason that customers shop on-line? Is it because they are seeking discounts or is it because they save time going to the shops, hunting down products or is it the convenience of home delivery? Logically, it's a mix of all three. But for many people, on-line shopping is about saving time. It's about convenience. Therefore the company must focus on delivery.

Unfortunately, it outsourced that to a company that provides very poor service. As a result, the reputation of the on-line company suffers.

The online company failed to learn, or if it learned failed to deal with, several specific issues :

First, the chargeable telephone number issued by the carrier is answered with a message saying "thank you for calling [company's] free telephone number.

Secondly, the "tracking" service says that a package is being delivered but this is not actually so. Worse, the tracking service reports "undelivered (absence)" when this is not so.

What actually happened, this author was told by a representative of the company, is that the delivery van is loaded. The driver chooses his route. Despite the fact that the company allocates delivery slots in four hour blocks from 09:00 to 21:00, the drivers do not stick to those blocks. At approx 19:00, the driver must return to base with packages that he has picked up during the day even though other packages have not been delivered. He does not then go out again with the packages he has not delivered. Instead he marks them as "absence." Therefore, there is a clear misrepresentation that he has attempted delivery but failed as a result of the consignee's non-presence.

When the driver decides to go to a destination, he calls and says he is on his way, putting the obligation to be present, at short notice, onto the customer.

As a result, a package may be marked in the tracking system as out for delivery for day after day but delivered at the whim of the driver, regardless of the inconvenience to the customer. There are other, less disturbing, signs of lack of attention by the carrier including the poor standard of English on their website which, as a primary means of communication, should be clear.

For the online company, the biggest headache is that, unless customers kick up a stink, its own reputation is sullied by the dishonesty and unreliability of that part of the chain that is actually customer facing.

So, why has the system failed? It's simply because no one fully beta tested it critically before implementing it, asking the basic questions "what can go wrong and how do we protect against that before it arises?"

Companies spend a fortune fixing problems that were foreseeable, if only they had asked the correct questions to begin with.

Worse, online system, in particular, are tested by people who design them or who live or work with those who design them. They must be tested by people who have no concept of what the site is or does or how it works.

An on-line money transfer agent's website is so complex that even finding the place to put a voucher for a free trial is so difficult that it is easily overlooked. Why would a designer put that behind a button instead of just placing a box for the code in the form? The system asks the user to choose whether to fund the transfer in one of a number of ways: but that choice is irrevocable so that, if there is a problem with, say, payment by card, it is not possible to substitute payment directly from a bank account. Worse, after the account is funded, the company sends an e-mail asking for due diligence information to be uploaded: why is this not part of the application process prior to depositing funds? Worse, now the company has the money and puts itself into a high-risk situation if the due diligence process fails.

E-bay wrote to its UK customers saying their account had been hacked and they they should change their password. This author, it turns out, had begun the process of opening an e-bay account in or about 2002 while investigating distribution channels for epublications. Having decided not to proceed, the application was abandoned but the account remained. Trying to change the password, it turns out that the account is blocked. So the password can't be changed. Solution: close the account. That, eBay say, is not possible (for which read we don't want to) because there is outstanding due diligence information. Send, they say, a copy of your passport. No, why would I give copies of official documents to a company I have no intention of dealing with, especially one whose subsidiary (PayPal) has in the past had such a tenuous grip on both law and truthfulness that I instructed my group of companies we were not to deal with them. So, I told eBay that so far as I was concerned, their account with me is closed, even if they do not close my account with them. Therefore any attempt by anyone to use the account they insist on keeping open will be an attempt by criminals and it is, therefore, entirely at their own risk.

Much of the credit for failure goes to those who demand that systems are followed inflexibly, as in the eBay case. Often, this means that customer service departments are set up not to deliver service (that, after all, is what the entire organisation is supposed to do) but to obstruct any attempt to obtain actual service on anything other than the equivalent of a vending machine's take it or leave it approach. And with a nebulous organisation, one that lives online or on the phone, you can't kick it when it takes your money and does not deliver the service.

Several years ago, the UK passport office went into catastrophic meltdown.

The official statement was that the passport office has been inundated with 300,000 applications beyond the norm. That's reactive: the question is why was that increase not anticipated? What has changed that has led to such an upsurge?

There are several hints, not the least of which is a change in process for the issue and renewal of passports for UK citizens resident overseas. Over recent years, there has been a series of changes each of which have reduced the quality of service and increased inconvenience and cost. First and best, there was the system to turn up at the local High Commission or Embassy with my old passport, fill in a form, hand over my photos and cash and pick up a new passport the next day. And I could get a big passport to accommodate all the stamps I collect on my business travels and, when it was full, I could get additional pages added for a small fee. Next, all passport applications for South East Asia were centralised in Hong Kong and passports had to be couriered at considerable expense. Worse, all passports were now smaller and additional pages were abolished. As a result, passports fill up quickly and instead of a ten year life have, to all intents and purposes, a three to five year life especially when visiting countries that, still, insist on affixing full page visas or passes. Fully aware that this system would soon fall down, I arranged for a second passport, with full permissions.

That's how I was able to get home from the UK when I needed to renew one of my passports: it was both full and expired (I kept it because even though it was full it had, until expiry, some valid passes).

Early this year, the system was changed again: now all passports have to be handled by The Passport Agency in the UK. Luckily, I was in the UK for two months so, with far more than the three weeks that the Passport Agency says renewals take, I sent it to the required address - Liverpool. Why, one has to ask, could I not send it to my local passport office in London where, if there was a problem, I could drop in? It's what I used to do when I lived in the UK and, in extremis, we could get a passport in about four hours. What happened? They were in such a mess that I left the UK on my second passport and several weeks later they couriered my replacement passport to me. At my cost. Then they told me they would refund the cost but, I've not found the refund in my account. Maybe I missed it.

Why does this matter to financial services businesses and to compliance and risk officers in particular.

Let's leave aside the obvious that an unhappy customer tells lots of people whereas a happy customer tells a few. Let's start with the point that the relationship between a customer and a business is fundamentally unstable and that the only way it becomes stable is for the customer to be encouraged to be loyal to the provider.

The balance of power in a financial services relationship is with the provider. That means that the customer is always in a position of weakness. Customers resent that they can do nothing to get the service they want or, in some cases, the service as advertised.

Long ago, it was drummed into me that the test of a management system is not how well it succeeds but how well it deals with failure. Things go wrong for all manner of reasons but rarely are they unpredictable. Mostly because it's because someone wanted to be too clever, or who had too much belief in themselves and their abilities. Mostly its because no one thought to assume that the river is flat and everything should relate to that.