Web Self-Service Management Principles and Business Case

By Gerry McGovern


Online is self-service






Constant observation, continuous improvement

Understanding customers through constant observation

Continuous improvement through testing

Business case for self-service

Balancing fixed cost and variable cost

Discovering the sweet spot for self-service

Finding the self-service sweet spot for Microsoft Legal intranet

The hidden costs of self-service

Channel shift, hidden demand and channel yo-yoing

Adapting in a complex world

Online is self-service

“Oh, we have a self-service section on our website,” some say. No you don’t. Your entire website is self-service. Your apps are self-service. Most of your digital ecosystem is self-service. “But isn’t self-service just about doing a transaction?” No. If someone is looking up a policy on the intranet, that’s self-service. If someone is trying to figure how to install a product on your website, that’s self-service. If someone wants to know what the weather will be like today, that’s self-service.

The principles of self-service design and management are:

  • Simplicity
  • Convenience
  • Control
  • Speed
  • Price
  • Customer observation, continuous improvement

Self-service is about people doing things for themselves rather than having others do things for them. On the Web, people self-serve by reading content, clicking links and using applications. The Web is all about self-service. Apps are all about self-service. The more technology-driven a society becomes the more self-service it becomes.

Self-service is part of an inexorable trend where people are being replaced by software, by content, by robots. Self-service means using something (content, applications) to fulfill your needs. Most buyer/seller relationships are not pure self-service. A vending machine needs to be refilled. Someone packs your order in Amazon. But the overall trend is towards self-service supported by content and apps.

The essence of the return-on-investment argument for self-service is that it reduces time. The organization spends less staffing self-service tasks and customers spend less time doing them.


Simplicity should be to self-service as chocolate is to joy or sadness is to taxes. Human-to-human service (the ‘opposite’ of self-service) is rich in flexibility, cues and empathy (unless you’re dealing with an ignorant service professional). Human interaction oils the path towards successful task completion. A self-service design must be very simple because it doesn’t have that human oil.

Imagine you’re a manager of a restaurant. You’re standing near the front door and you see an old lady approach. If you’re any good at your job you will open the door with a smile. However, that same old lady could be squinting at the tiny, grey text on your website and you will in all likelihood not even be aware she exists.

It’s hard to say to a webpage: “I don’t understand you.” It’s hard to say to an app: “Are you sure this is the right direction?” (I mean, you can say what you want to a webpage, scream at it if you want, but it’s not likely to respond.) A friendly Human Resources manager may say to you: Don’t worry about reading that section. It’s not relevant to you. And this section here, this is what it actually means.” They are bringing years of experience to bear on the human-to-human interaction in which they are engaged.

A good doctor can make a serious and complex situation reasonably simple for a patient. They can bring empathy and understanding and thus reduce stress and confusion. They can adapt to the specific demands of the here and now. They might meet eight patients in one day and deal with each one of them slightly differently, even though they are dealing with the same diagnosis.

That’s very hard to do in self-service. Sometimes, it’s impossible. There are limits to self-service. Not every service is self-serviceable. Some services are actually cheaper to deliver on a human-to-human basis (as odd as that seems).

Complexity is a self-service killer. When faced with a complex self-service environment most people will either:

  1. give up,
  2. use another channel, and/or
  3. scream, and try again

It’s very difficult to design and manage a simple self-service environment. Very difficult. Take content, for example. The exact words you choose are incredibly important in order for people to:

  1. Find what they are looking for
  2. Understand what they have found (assuming they have found what they were looking for)
  3. Find what they actually need, rather than what they thought they needed and were looking for

Sound difficult?

When you go into a self-service restaurant you don’t want to search for the menu. When you go into an expensive restaurant, you often have to wait for your waiter to give you the menu and pronounce for you the complicated French-sounding words and explain them, while you nod and give the impression that you understood the menu from the get-go. It’s a little bit of a ritual, adding to the atmosphere and your education. (That must be nice because it’s simply unpronounceable.)

Self-service is a different beast entirely. The stomach is rumbling and the pocket doesn’t have much cash. Once you find a self-service menu you tend to scan it quickly. If it has too many choices or the words aren’t simple, chances are you’ll get annoyed and either say “give me a Big Mac meal, please” or leave.

A lot of content that ends up on webpages or in applications is either:

  1. Not written specifically for self-service. It has been written for experts like the Human Resources manager, who would then bring ordinary employees through it, or
  2. Written with self-service vaguely in mind, but has not been properly tested.

A huge amount of stuff that ends up on the Web has been created either deliberately or implicitly for human-to-human service. The human resources policies really do need a human resources expert to take you through them. The installation guide will make perfect sense once you have the engineer who designed the system sitting beside you. The description of the course truly comes to life when you have the professor responsible for it on the other end of the phone.

It’s easy for organizations to create complexity for their customers because it makes things simpler for themselves. There’s a real tension: making it easier for the customer invariably makes it harder for the organization. Years ago HSBC bank in Hong Kong had a form on its website for those interested in mortgages. It worked well from an organizational point of view:

  1. It collected all the information needed to qualify a lead.
  2. It created a nice data picture of the applicant.
  3. It integrated very well with internal systems.

In order to enquire about getting a mortgage, a customer must fill in 17 fields.


There was one problem with the form: it only generated 2 enquiries a week.

They decided to simplify the form. This is what they came up with:


Internally, there was resistance.

  1. It provided a very basic data picture of the applicant.
  2. It didn’t integrate well with internal systems.
  3. Staff felt they would be wasting time on applications from “Mickey Mouse” because it was such an easy form to fill out.

180 enquiries a week with the new form. Yes, there were enquiries from Mickey Mouse. However, in the previous quarter with the more complex form, less than $1 million in new mortgages was generated from leads that came from the form. In the first quarter with the simplified form, $20 million in new mortgages was generated. Simplicity for the customer creates complexity for the organization, but the value generated by more customer business can more than compensate for that extra organizational complexity.

“Keep in mind that, in general, customers’ use of Web self-service is infrequent,” an Oracle white paper stated in 2012. This is equally true for employees. Many older systems were full of complexity. People needed extensive training on them and only after they used them regularly did they find a way to be relatively efficient.

For example, suppose you rang the travel agent who was using the booking system all day. You used to have a secretary or administrator in your department who regularly booked meeting rooms, ordered food, etc. Now it’s self-service and you occasionally need to do those kinds of things yourself. You want these apps to be really simple. You certainly don’t want to have to take training to use them. (And even if you did you’d have forgotten what you learned the next time you needed to make a booking.)

There’s a hidden cost of self-service within organizations. Before, it was administrators and secretaries who were carrying out a lot of tasks such as booking flights, reserving meeting rooms, etc. Because they were really experienced and familiar with the system, they could book a meeting room in 3 minutes. However, now managers or product designers are expected to book their own meeting rooms. If the new self-service is really working then they should be able to book the meeting room in 2 minutes. However, it often takes them 6 minutes, twice the time of the administrator, because they’re unfamiliar with the system and it’s poorly designed. That’s more expensive time too; it’s managers’ time, not administrators’ time. That’s not productive. That’s not cost-effective.

A lot of organizations think they’re saving money by getting rid of people and replacing them with technology. That’s often true, but not always. “Toyota has found that the race to reduce the human element can end up making processes less efficient,” Max Nisen wrote for QUARTZ in April 2014. In just one example, people replacing robots resulted in a 10% reduction in waste for crankshaft production, and also shortened the production line. “We cannot simply depend on the machines that only repeat the same task over and over again,” project lead Mitsuru Kawai told Bloomberg. “To be the master of the machine, you have to have the knowledge and the skills to teach the machine.”

Technology is not god. We must constantly test so as to be sure that the technology and content we are using is actually making things simpler and better. Because complexity kills self-service. It takes away all its core advantages and strengths. Complexity online involves large, badly organized websites with poor quality, out-of-date content and that champion of complexity—confusing menus and links.


If you want to maximize self-service uptake then you must make it more convenient for the customer to use compared with other channel options. This is related to but not exactly the same as simplicity.

Renewing your car tax by queuing in some office is very time-consuming and annoying. If you can do it online, even if it’s not that simple, it still saves you lots of time and hassle. But if you think about filling out a complicated HR form, it’s often easier to just pick up the phone and talk to someone.

Perceived convenience or inconvenience is essential to understand here. Convenience is relative but then again, so is simplicity. Lots of people now want an ‘Amazon experience’ whether they are engaging with a public website or an intranet. What they would have regarded as convenient five years ago, today they regard as awkward and inconvenient.

To get someone to shift channels (from say the phone to the Web) you must prove that the self-service alternative is a lot more convenient, not just a little. Sometimes, that will require special training or incentives, because many people simply don’t trust technology to be simple.

I’m one of them. Even after the self-service kiosks arrived at airports I stood in the queue because I just didn’t trust the machine. One morning, a nice gentleman tapped me on the shoulder and offered to bring me through the process. He said that if it didn’t work, he’d immediately bring me up to the top of the queue. It did work and I’ve used it ever since.


People want more control today and that’s one of the greatest things the Web has given us. Control is addictive. One of the factors that cause addiction to gambling is when the gambler believes they have some control over the outcome. That’s why gambling addicts often have superstitious behaviour. We love control – we get addicted to it! On the Web, people feel more in control of their lives and their purchase decisions. They feel less manipulated by clever marketing and advertising.

In a supermarket, many people prefer to pick their own loose fruit, instead of having someone else pick them for them, or buy them in packs. Certain farms allow people to come and pick their own vegetables and fruit. It’s a nice feeling of control to be able to choose your own seat on a plane rather than have someone else choose it for you.

“Where customer relationships were once defined through marketing,” Cale Thompson wrote for Fast Company in 2013, “today’s customers increasingly demand control over the offerings themselves and may rebel if what they experience doesn’t match what marketing promised.”

Zipcar, a pioneer in short-term car rental in North America, has focused on transparency and giving back more control to the customer. “Imagine picking out a car via smartphone while waiting for your luggage, then walking to the rental lot and driving off. There’s no need for long lines, insurance forms, and sales associates trying to upsell you on the spot,” Thompson writes. That would be a dream. That’s convenience and control.

If you’re dealing with a medical professional today, you don’t simply accept their word. You research the condition. You check up on the medication that has been recommended. The next time you visit, you may have another medication in mind, and you will ask your doctor if you should be taking it instead.

People go online to do research in order to make better and more informed decisions. They want transparency. They want to see essential information such as price. There are many examples of the benefits of transparency but River Pools, a swimming pool manufacturer in Virginia, is one of the best. When the recession hit in 2007, people stopped buying pools. In desperation, River Pools slashed their advertising budget by 90% and tried a radical strategy. They told people who visited their website how much a pool cost.

“As a result,” according to a New York Times article, “River Pools has recovered to exceed its peak pre-2007 revenue.” How did you save your company?” the Times asked co-owner, Marcus Sheridan. “I just started thinking more about the way I use the Internet. Most of the time when I type in a search, I’m looking for an answer to a specific question. The problem in my industry, and a lot of industries, is you don’t get a lot of great search results because most businesses don’t want to give answers; they want to talk about their company. So I realized that if I was willing to answer all these questions that people have about fiberglass pools, we might have a chance to pull this out.”

Sometimes, there is a trade-off between control and simplicity. Early versions of software are often quite simple but over time they can become more complex, partly because they are trying to deliver more control. One customer’s control could be another customer’s complexity. A feature that might empower someone who is experienced with a product, might confuse a newbie.

Some believe that increasing the quantity of information available to people increases control and transparency. This can be true but giving people lots and lots of content about a medical disease or condition, for example, might leave them more confused, not less.

Words are the building blocks of the online world and there are only so many words to go around. So, we get the concept of ‘word overlap’ where the same words are used for many different tasks, and/or certain words can have a very different meaning depending on the task a person is trying to do.

Many technology company websites like to have a section called ‘Solutions’. This is where they describe how their products and services combine to offer an overall solution to the customer. However, when we gave customers support-type tasks, such as troubleshooting a problem, they often clicked on this ‘Solutions’ link. Why? Because they were looking for a ‘solution’ to their troubleshooting problem.

The Microsoft Excel team wanted to give more control to mathematicians by providing descriptions of various mathematical functions. The function pages had names like ‘IMSUM, Print, Areas.’ Far more people needed to learn how to sum a number or print a page, but a significant number of them ended up very confused as they browsed function pages. What was happening here was that a top task (sum a number) was being negatively impacted on by a tiny task (IMSUM function).

The solution? They deleted all the individual function pages and placed them under a single page called “Maths Functions.” The mathematicians could still find them but they weren’t getting in the way of people who just wanted to learn how to print a page or sum a number.

When you are adding new features or content always ask how it impacts what’s already there. In particular, don’t allow tiny task features and content to negatively impact the performance of top tasks.


Speed is critical for self-service success. You may pay to wait for that wonderful, slowly cooked food in a plush restaurant, but you certainly don’t pay to wait in McDonalds. It is, after all, part of the ‘fast’ food industry.

“The amount of time that consumers are spending waiting in lines at fast-food drive-thru windows is getting longer, not shorter, mostly due to the growing complexity of new products that the major fast-food chains are selling,” Bruce Horovitz, wrote for USA Today in 2013.

Many US fast food chains do 60-70% of their business with drive-thru, so it’s a very important channel and the time it takes to fulfill an order is a key metric. What’s causing the extra waiting? Menu bloat. The industry is constantly releasing more premium and complex products. It’s a very difficult problem to address because on the one hand the consumer is demanding more but is not ready to wait. Speed isn’t always everything, of course. Accuracy and hotness matter too. Unfortunately, complexity is also resulting in less accurate and cold orders being delivered.

Google is the king of speed. Larry Page, founder of Google, is obsessed with speed. “Before Google launched Gmail in 2004, its creator, Paul Buchheit, brought it to Page’s open cubicle office for a review,” Nicholas Carlson wrote for Business Inside in 2014. “As Buchheit called the program up on Page’s computer, the boss made a face.
“It’s too slow,” Page said.
Buchheit disagreed. “It was loading just fine”, he said.
“No”, Page insisted. “It had taken a full 600 milliseconds for the page to load.”
“You can’t know that,” Buchheit said.
But when he got back to his office, he looked up the server logs. It had taken exactly 600 milliseconds for Gmail to load.

That’s the culture of the Internet. That’s the culture of speed. As a result of observing customer behavior, Google has found that any delay of greater than 500 milliseconds (half-a-second) has a negative impact on customer behavior. People don’t consciously notice such delays but they ‘feel’ the slowness and will avoid, if possible, an environment that feels slow.

Bing, in its own studies, found the exact same sort of negative behavior occurring as a result of a delay of 500 milliseconds or more. In one study it found that if there was a 3 second delay beyond the average for search results to appear, there was a consequent 4% drop in advertising revenue from that page.

“We had a similar experience at Amazon.com,” Greg Linden of Amazon states. “In A/B tests, we tried delaying the page in increments of 100 milliseconds and found that even very small delays would result in substantial and costly drops in revenue.”

  • A one second delay in page load times, according to Strangeloop, leads to a 7% reduction in conversions.
  • You have approximately 7 seconds to orient a customer once they arrive at your page, according to Marketing Experiments.
  • Walmart found that there is a sharp decline in conversion when the average load time for a page rises from 1 to 4 seconds.
  • Speed is paramount for European online shoppers, according to research carried out by Loudhouse in 2013, with almost half of customers surveyed claiming they would abandon a purchase if the webpage loaded too slowly.
  • According to a 2013 Maxymiser study, 22 percent of respondents are willing to wait no more than 3 seconds for pages to load on bank websites. “What’s most interesting about these findings is that what consumers say they want doesn’t match their actions,” Maxymiser states. “It is a widely known fact from research by Akamai and Forrester that two seconds is the threshold of acceptability for ecommerce web page response times.”
  • People make subconscious judgments about a page after a mere 50 milliseconds, according to research reported by Fidelity Investments.


People have an expectation that if they’re serving themselves they get some sort of reward or payback for the effort they are putting in and the effort they are saving the organization. Of course, this may be balanced by the convenience and control factor. If self-service is so much more convenient or gives you much more control, then the customer might even be willing to pay more, but that would be an exceptional situation.

There is an expectation at play here: “I’m doing all the work myself. I’m saving you money. What’s in it for me?” That’s why McDonalds is cheap. That’s why Amazon is cheap. That’s why Walmart is cheap. That’s why IKEA is cheap. If you want to maximize self-service uptake then you make it cheaper, faster and simpler.

Customer observation, continuous improvement

Understanding customers through constant observation

Self-service design is really hard because it requires an understanding of the gut instinct responses of customers; something of which they are not always even aware. So, how do you design for self-service?

  1. Constant observation of customer behavior
  2. Continuous incremental improvements based on testing

The worst possible way to design for online self-service is to have five smart people in a room drinking lattes. The longer you leave them the worse the design becomes. (The next worst way is to have 15 customers in a room drinking lattes, because what they say is rarely what they actually do.) The five smart people go with their gut or their experience, but again and again we have found that the organization’s gut instinct is rarely the same as the customer’s gut instinct.

Sam Walton founded Walmart. When his family went on vacation they always tried to choose some rural, out-of-the-way destination. They knew from bitter experience that if it was anywhere near a town or city, then Sam would disappear for long periods. They’d find him scouting around the local supermarkets, watching, observing. Once they found him on his knees as he tried to figure out the optimal height of shelves. That’s the life of the self-service pioneer.

Many web teams think of flesh-and-blood customers (if they think of them at all) as vampires think of crosses and garlic. Something most definitely to be avoided. Web teams love to pore over code, content, graphics, or ‘user’ stats. Anything but actual flesh-and-blood customers themselves.

You can’t do self-service well if you don’t continuously, relentlessly observe your customers. It is by far the biggest flaw and blind spot of web teams.

When Ray Kroc realized that the McDonald brothers were doing something interesting with their restaurant, he got in his car and headed out to San Bernardino. But he didn’t go straight in and talk to the brothers. He parked on a hill and for two days watched intently the comings and goings of customers. Ray then went to the brothers and made them an offer. And the rest, as they say, is obesity.

I once had a conversation with a McDonalds manager who told me that one day she was in her office when her superior called. She picked up the phone and the first thing she was asked was: “What are you doing in the office? Why aren’t you out in the restaurant?”

All successful self-service environments—whether they are offline or online–relentlessly strive to better understand their customers’ behavior. It is quite simply not enough to ask customers what they think and what they want. Most of the time they just won’t tell you what they want, or they’ll fictionalize or romanticize their wants.

A number of years ago British Airways introduced a small fridge for first and business class customers on long distance flights. The idea was that when customers would wake up, feeling a little hungry, they could go to the fridge and have a snack, rather than having to ask an attendant. They did their research! They asked customers in Tokyo, New York, London. And what did their customers tell them? That they wanted light salads, greens, stuff that’s good for you.

When they were loading the fridge for the first time on a flight from London to LA, a flight attendant walked by.

“What are you doing?” she asked.
“It’s for our customers,” they said and explained the rationale.
“They asked for what?” the attendant said incredulously.
“Light salads, greens, healthy things …” She laughed and walked away.
A few minutes later she came back with a bunch of chocolates. “Put these in the fridge, please,” she said. “I know, I know, but trust me. I’ve been dealing with them for 10 years and I know what they really want when they wake up.”
So they put some chocolates in the fridge. And when the flight landed, all the chocolates were gone while the salads were left untouched.

You simply can’t trust what customers tell you most of the time. You have to find other ways of discovering what they really want. Most of the time that involves observing what they are actually doing.

Continuous improvement through testing

The magic sauce of self-service management and design is to take lots of customer observation and mix it vigorously with continuous improvement. The cultures that Sam Walton and Ray Kroc developed were based on continuous testing and tweaking, a constant refinement and evolution.

The more complex the world becomes the harder it is to have effective plans and strategies, because plans and strategies are all about predicting and / or controlling the future. Not easy anymore. Therefore, organizations need to become much more adaptable, nimble and flexible.

Self-service environments are designed in the lab of human interaction, aka the Web. It’s an online world and if you want to, you can have a front seat observing the comings and goings of people in a way that was never before possible.

On the Web, people leave a trail, whether they want to or not. Those who develop an understanding of the clues left on the trail will design the best and most intuitive self-service environments. Data is the new oil. Modern industries run on customer behavior data.

You don’t have to be there to be there. Remote testing and observation is revolutionizing the way we can understand our customers. Essentially, remote testing involves observing the customer’s screen and being able to hear their voice (although being able to hear their voice is not always essential). In understanding online behavior, remote testing has many advantages over traditional lab-based testing. It is:

  1. Faster: You can set up tests much more quickly and more often. It’s a lot easier for someone to give you 15 minutes or 1 hour of their time online than for them to spend a morning visiting your lab. Because of this, remote testing can become part of the work-week, not something that is done occasionally. That’s a huge advantage because observation can become part of management culture, something that must happen in order to have quality self-service environments.
  2. Cheaper: ‘We can’t afford to test,’ was often a refrain. Well, no excuses now. The cost of setting up a remote test is low. Oh, yes, it does take time. But there is nothing more important a manager or web team can do than spend time observing their customers. Ray Kroc understood this. Sam Walton understood this. Starbucks, Amazon, Google understand, and every successful self-service entity understands.
  3. Better: ‘Now, please sit down at that computer you’ve never used before. And don’t mind me sitting beside you with my notepad. And forget about that camera that is broadcasting your every move to the web team in the next room. And ignore those strong lights. So, ready. Just act natural. Pretend you’re at home or in your office.’ You get better, more real and more natural behavior if the person is actually at home or in their office, using their own computer with nobody sitting beside them with a notepad.

On a Monday you come up with a hypothesis of something you’d like to change on the website based on customer behavior data analysis. It could—and generally should be—a very small change, like changing the text in a link. On Tuesday you make the change live and then on Wednesday you test with real people to see if the change had its desired effect. Not quite? Okay, edit the link and observe again. A continuous, never-ending process.

You will end up testing words more than anything else. The Obama Presidential campaigns were acknowledged masters of using the Web.


In the 2008 campaign they rigorously tested the words they used for getting people to donate funds. They found, for example (as can be seen in the preceding table) that if you had not given money, then “donate” worked well, but if you had given money at least once then “contribute” worked better.

In the 2012 campaign, they made their webpages download 60% faster and this resulted in a 14% increase in donation conversions. According to Kyle Rush, former Deputy Director of Web Development for the Obama 2012 campaign, “By the end of the campaign our 240 a/b tests lifted the donation conversion rate by 49%!” And how much did they raise online? $250 million. As another Obama campaigner put it, “The time of guys sitting in a back room smoking cigars is over.” This is the age of observation, of analytics, of continuous improvement, of decisions based on data.

The Web is the ultimate laboratory of human behavior. Dive in. This world is far too complex for any of us to predict the future, to write the perfect sentence, to design the perfect interface. We do our best to get started and then we test and tweak, test and tweak. That’s how we maximize value. That’s how we maximize customer satisfaction and loyalty.

Business case for self-service

The worst thing you could possibly do is make everything self-service. ‘Let’s put everything online’ has a lovely ring to it. You hear it in government a lot. We publish everything because we’re transparent and we want to serve citizens. It’s a noble ideal but unfortunately it usually ends up with an unmanageable, unfindable, unusable online presence.

Self-service doesn’t work for everything. There are services that are much more effectively delivered over the phone or in face-to-face situations. Services delivered this way cost less and lead to greater customer satisfaction. That might sound counterintuitive but it’s true.

Self-service design is really hard and expensive to do right. It takes lots of testing and observing in order to optimize. Online, customers are ruthless. If you don’t deliver simplicity on their terms in seconds, they leave. If they have to do business with you then they’ll send you an email, make a call, or walk into your office.

Some services are very complex. I talked to a manager of a council / municipality who told me that services connected with planning permission were very hard to deliver online. No matter how many times they rewrote the instructions they kept getting applications full of errors and misunderstanding. So, they ended up deleting most of the content and simply requesting that people phone them for a 30-minute discussion. Much better applications were received, saving everyone lots of time and money in the longer term.

Balancing fixed cost and variable cost

When deciding on whether to choose self-service or not there are two types of costs that need to be calculated:

  1. Fixed cost
  2. Variable cost

Self-service has high fixed costs and low variable costs. Face-to-face service has low fixed costs and high variable costs.

Creating a website or app is cheap. Hire a student. Creating and managing a website or app that is really simple to use is hard and expensive. Simplicity just doesn’t come cheap. It takes lots of effort to create a simple self-service environment and it takes just as much effort to maintain and evolve it. But once you have created that simple self-service environment then the cost per task completion (variable cost) tends to be relatively low. Whether a room booking system has to deal with 100 or 150 bookings per hour doesn’t change the costs very much. Now, if it had to deal with 10,000 per hour, then yes, it would need lots more bandwidth and processing capacity, but the basic variable cost per transaction tends to be quite small. In general, the more task completions (transactions) the better the business case is for self-service. As self-service scales, the cost per task completion tend to decline. So, the cost per task completion if you are processing 10,000 tasks per hour tends to be less than if you are processing 100. Self-service is very scalable.

It’s the opposite with face-to-face service. The variable cost is high and constant because we’re dealing with two humans interacting. If it takes 10 minutes to deal with customer A, then it will take roughly 10 minutes to deal with customer B. The variable cost is that 10 minutes of time repeated every time you have an interaction between a staff member and a customer. In general, the higher the demand from customers the worse the business case for face-to-face becomes. For low demand, face-to-face often does better because the fixed cost is not so high when compared to self-service. From a fixed cost point of view the physical infrastructure that staff work with (offices, phones, etc.) are often already accounted for and/or can be spread over a number of functions and activities.

The following is a hypothetical situation, although the variable costs are real. (They come from Socitm, a UK organization that consults for local government.)


As we can see, the variable cost of a web task completion / interaction is low at 27 pence per interaction, while for face-to-face it is much higher at £9.34 per interaction. Given only these figures you’d automatically say: Let’s make everything self-service. We’ll save lots of money. Ah, if only it were that simple.

The fixed cost needs to be accounted for too. The web environment (content, software, bandwidth, usability, ongoing management, etc.) needs to be budgeted for. The web must work and to work it must be simple. We know that simple costs and very simple costs even more. What happens if you have a complex web environment? Task failure. Dissatisfied customers who either go to a competitor or ring you on the phone or visit your office.

Let’s say John the customer tries to complete his task on the website and fails, and is forced to ring you. What’s the cost now? It’s not just £3.76 (see preceding chart). It’s £4.03 (the web cost of 27p plus the phone cost of £3.76). So, basically, self-service only works if it works. Poor self-service adds to the organizational cost base and reduces customer satisfaction – a lose-lose situation.

We’re estimating in the previous table £5,000 of fixed cost for face-to-face and £40,000 to develop the web environment. What does this mean? That for low volume tasks, face-to-face has a better business case, as can be seen from the following graph.


The preceding chart tells us that:

  1. Up until about 1,800 task completions, face-to-face has the best business case.
  2. From about 1,800 to about 7,300 task completions phone has the best business case.
  3. After 7,300 task completions, the Web has the best business case.

Discovering the sweet spot for self-service

In deciding what does and doesn’t work for self-service, the following factors need to be taken into account:

  1. Online demand: Self-service thrives on high demand: Are there lots of people who want to complete the task online? There might be lots of people who want to or need to complete the task but might not want to do it online because they feel it’s not as convenient, or they feel vulnerable in some way, or they think it’s just too complex. The larger the volume of task completions the better it is for self-service.
  2. Complexity: Self-service thrives on simple tasks. The more complex the tasks the more and more complex the content you will need to create. Your app will also become more complex with more features and more steps in the process of completing the task. And remember, the complexity that you add does not simply stay confined to the task to which you added it. It bleeds into the navigation and search and it eats up valuable time that your web team could be spending improving the performance of top tasks.
  3. Costs:
    1. Staff costs: How much does is it currently cost to deliver this task / service? How much do staff cost and how long does it take to help the customer complete a task? The more expensive the staff are and the longer it’s taking, the more potential for self-service benefits. (Although the cost of the staff and the length of time they need to spend helping customers complete the task in question are also likely a factor of the complexity of the task.)
    2. Online costs: How expensive is it to create a website, content and apps for the task? This needs to be balanced against the staff and other costs connected with phone and face-to-face.

If we take the three preceding factors into account then we can chart a sweet spot for self-service, phone and face-to-face.


The preceding chart shows us that the sweet spot for self-service is where:

  1. The task is simple to complete
  2. There is high demand to do it online
  3. Current staffing costs to deliver the task by phone or face-to-face is high

The sweet spot for phone is where:

  1. The task if fairly complicated
  2. There is just reasonable online demand
  3. The staffing costs are reasonable

The sweet spot for face-to-face is where:

  1. The task is quite complex and involved
  2. There is low online demand
  3. Staffing costs are low

Obviously, it’s not possible to find ‘sweet spot’ tasks every time. There are trade-offs for sure. It may well be that the task gets broken up into sub-tasks. You can have a phone call to get an overview of the application process, then you can submit your application online, then you can have a face-to-face interview to discuss the decision. It’s a balancing act of costs, complexity and demand.

Here’s what happens if you bring tasks that are best suited to face-to-face online:

  1. They disrupt the tasks that are suited to self-service. How do they do that?
    1. They make them harder to find through search. Face-to-face tasks tend to have lots of content, thus reducing the effectiveness of search. This is particularly the case where word overlap occurs. (Tiny tasks often contain the same keywords as top tasks.
    2. They make the navigation more difficult and confusing.
    3. They eat the web team’s time. Many teams have told me that they can’t focus on making things better because they are so busy publishing.
  2. They fail. They don’t achieve anything except frustrated customers and increased costs.

Finding the self-service sweet spot for Microsoft Legal intranet

LCAWeb is the part of the Microsoft intranet dealing with legal issues such as contracts, immigration, copyright, compliance, etc. It wasn’t being used enough by employees and legal staff were complaining that they were constantly being asked questions whose answers were already published on LCAWeb. As is usual for intranets, there was a lack of resources to make things better. They needed to prove the business case for the intranet.self-service-ms-employee-7
As shown in the preceding chart, the average cost for a legal professional to answer an employee question over the phone was $33. If this was done by email, then the cost was $10. For employees completing tasks using the intranet the cost was $1—33 times cheaper than over the phone.

But the intranet wasn’t working. Staff would visit, not find what they were looking for, then email or phone. The intranet was adding to costs, not reducing them. A major reason why the intranet wasn’t working was because it was using an organization-centric navigation.

For example, if you were creating a brochure and you wanted to find out what fictitious names you could use (like John Brown), it was hard to find. Legally speaking, fictitious names are a type of trademark, and thus they were placed under the link “Trademarks”. But in testing only 16% of participants found them there. So, the team did a bit more research in order to understand how staff thought. They discovered that staff looked for fictious names as part of marketing activities. When they placed them under marketing, 100% of those they tested were successful.


There was some resistance from the legal experts because they felt that legally speaking they should be placed under Trademarks. However, once they saw the test results and saw how this would reduce phone calls, they were convinced (at least for a while).

The hidden costs of self-service

When you look at the following table it would look like a no-brainer, a slam-dunk to make as many transactions as possible self-service. It’s so much cheaper.


However, continuous improvement and hidden demand are hidden costs of self-service. In my experience of working with the Web since the mid-Nineties, most websites or apps launch with about 60% task efficiency. In other words, only 60% of customers can complete their tasks and for those who do, most spend much more time than they should. The Web is never finished. You cannot launch and leave. If you want to maximize value and customer satisfaction you must continuously improve. Most organizations simply don’t budget for continuous improvement.

It’s a shame because they are losing out on so much value. In 2012, the Technology Services Industry Association (TSIA) estimated that the average cost for a self-service (Web) task completion was 6 cents. Yes, you read it correctly: 6 cents. The average cost if a technical service rep had to go and physically visit a client was $797. Big difference. However, the TSIA also presented the following chart.


Successful self-service visits to websites had declined from a lowly figure of 48% in 2003 to an abysmal 39% in 2011. Even though self-service had a fabulous business case, self-service performance was getting worse. Why? Because the 6 cents per task completion reflected significant underfunding. The technology websites were very poorly funded and thus they were getting bigger and worse.

  • The cost of a face-to-face visit was 13,283 times more than the cost of self-service.
  • If 10 times more was spent per self-service task completion it would still only amount to 60 cents.
  • If 100 times more was spent that would only amount to $6.

Lack of proper funding for self-service hurts everybody. Let’s say you’re a technology company and your self-service fails customers 40% of the time, and they are thus forced to phone you. Let’s say you have 10,000 customer task requests a month. Using the phone cost of $162 versus the self-service cost of 6 cents from the earlier table:

    1. If you had 100% success rate online then the cost per month would be $600 (10,000 X 6 cent)
    2. If you have 40% failure rate for self-service, then the total cost is $648,600. This figure is made up of:
      1. 6,000 people who succeed at self-service, costing $360.
      2. 4,000 people who fail at self-service, costing $240.
      3. 4,000 of these frustrated customers now must call you, costing $648,000.

Channel shift, hidden demand and channel yo-yoing

In the perfect world, self-service causes channel shift. People move away from the more expensive channels of phone and face-to-face and use the Web. Everyone is happy, everyone saves time and money.

Unfortunately, what often happens is channel yo-yoing, where people are forced to bounce between several channels in order to complete their task. Everybody loses. The customer is dissatisfied and the organization loses money and customers.

However, even where self-service is successful it may unearth hidden or latent demand. Before the Web, you waited for your bank statement to arrive in order to check your bank balance, and only rang up in exceptional circumstances. Now with your online banking you find yourself checking you balance regularly.

In October 2008, there were 2,202 calls to South Tyneside Council for waste and recycling tasks. There were 203 web interactions. The council decided to greatly simplify the waste and recycling webpages. By April 2009, there were 1,946 phone calls and 3,922 web interactions. So, the number of web interactions exploded but the number of phone calls didn’t drop all that much. The simpler website had tapped into a latent, hidden demand.

I dealt with a bank once that found that when it simplified its website it stopped getting calls about basic questions. It was getting less calls but they were taking longer, as they were more complex. What it found over time was that many of these ‘support’ calls were turning into sales opportunities, as customers realized they needed more sophisticated services from the bank.

Hidden demand, by its nature, is difficult to predict. It is equally difficult to quantity its cost and benefit. Some customers might become information addicts, checking the stock price or weather report every hour. This sort of customer behavior may become a net cost. Other customers may start engaging more with you, and as a result buy more / do more with you.

Adapting in a complex world

The more complex the world the harder it is to predict the future. Thus, the more complex things become the more adaptable we must become. We shouldn’t create things today so that they are built to last, but rather so that they are built to change. Only fools predict the future. The wise ones have a clear vision of where they want to get to and then prepare for all eventualities. They are nimble, flexible and constantly refining based on feedback from their environment.

One of the best ways to develop flexibility and adaptability is to look outwards towards the customer, rather than inwards towards the organization. Looking in encourages tribal / organizational, rigid thinking and behavior. Forcing yourself to think like your customer is like going to a mental gym. It’s not easy. It strains you. It makes you sweat. But the rewards are unquestionable in a world that is increasingly dominated by demanding and skeptical customers.

Self-service is the core underlying trend that drives the online world. People go online to do things for themselves, to serve themselves. They behave in a highly impatient way and demand that things are simple, fast and convenient. They like the feeling of being in control, and if they can get a good deal as they serve themselves, all the better.

Excellence in online self-service is about continuous improvement based on observed customer behavior. Smart people drinking lattes opining about customer behavior doesn’t cut it anymore. We need the data on customer behavior. We need to relentlessly focus outwards in a cool and objective way. In the online self-service world the customer isn’t king; they’re dictator.

In self-service management remember one thing: the ‘self’ is not you.

Download PDF version of Web Self-Service Management Principles and Business Case (PDF 733 KB)

Email Customer Carewords