Why client+agency engagements are so often failures, but it’s a secret

Whilst I’ve never worked agency-side, I have a perspective from client-side on why client+creative agency engagements are so often not as productive as they should be. I’ve worked with many creative agencies – as well as other suppliers, and there are some common themes that I’ve identified. There are issues on both sides.

In this post, I’m generalising. I’m sure there are some great supplier managers, and some great agencies, that really make things work. I have seen moments of sanity impinge on this process, but not as often as one would expect.

Senior managers drive it

Taking on an agency is usually driven by senior management. That’s not a surprise in itself. They are the ones carrying the can for effective results, and they have the budgets to pay for the work. The problem is that they often don’t involve the workers, don’t always understand the actual problem they are trying to solve, don’t know how to set the right goals… erm, etc.

Whilst the worker bees are, to mix metaphors, beavering away, they find that all of a sudden  there’s this agency that’s been taken on to do something or other, and the worker bees may or may not be expected to engage in the exercise. It’s not always clear.

Often, whatever the deliverable is, it’s the worker bees who are expected to manage, operate, and update it once the engagement is over. But there’s been little engagement or input from the client bees, so they feel no emotional involvement with the solution, and may even think that it’s the wrong answer.

This isn’t just an issue that applies to agency engagement – too often senior management will think that they know enough to drive a solution without talking to the people on the ground.

At the end of the engagement, it’s clearly not in the interests of the commissioning manager to say ‘Well that didn’t work’. If they do, then they typically have to blame the agency. Unless you have someone who is remarkably non-defensive and open to learning, they aren’t going to admit that the terms of engagement and processes weren’t right from the start. The result is that the same mistakes are made next time.


  • Owning managers should consult their own staff to shape an engagement
  • Client staff should be involved throughout the engagement at all levels

Agencies load up the client team

Particularly with larger agencies, there are going to be a cast of players whose roles are never really quite clear, but who charge high fees for each hour worked. There’s a relationship manager, a relationship director, a project manager, a creative lead, an interaction designer, etc etc, and when there’s a meeting some turn up, and some don’t, and the client doesn’t get a lot of say in it. I have serious trouble believing that all these roles are either needed, and need to be engaged at the level they are. It’s certainly not Lean, and it costs a lot of money.


  • Agencies need to be transparent about roles
  • Agencies need to clearly justify roles and billable hours
  • Everyone client side should understand who’s in the meeting, and why, not just the owning manager

Objectives aren’t clear, and most aren’t measured

I think it would be fair to pay an agency on results (or at least, have an element of it). It’s likely some do that. But it’s hard if the client organisation isn’t clear itself about hard business objectives that need to be met, and the role the agency can play. Obviously there’s a difficulty for an agency if the client refuses to implement their recommendations, which is why the mixed compensation model may be best. It’ll depend on the brief.

Some time ago I was strongly dissatisfied with what a marketing agency were doing for us. I talked to one of our internal marketeers (who agreed with me), and who arranged a meeting at the agency offices for the entire team working on our account – about 12 people I think. We had a discussion. At one point I asked how they knew if what they were doing for us helped us meet commercial goals. There was a slight pause of astonishment that such a question should be asked, and then it was carefully explained to me that the people in the room were all experts in their roles, and there was a large amount of cumulative experience brought to bear on the work. That was it. No reference at all to measurement, testing, metrics, anything. And it seems that both sides were ok with that.

A particular issue that I’ve come across is that I’ve often seen agencies obsessing over details of visual design, whilst completely missing the point that the interaction design was poor. Without a clear focus on commercial goals, and the right balance of skills on the project, there’s a risk of ending up with a visually delightful, commercially useless solution.

As a business, you need to be able to define quality standards for the work that an agency does, be able to police those standards, and have a process that kicks in if they are not met. Take note, anyone who plans IT outsourcing.


  • The client needs to be clear about the business goals for the engagement
  • Business goals should be written into the contract and impact remuneration
  • How to measure the business goals may need to be planned, and measure put in place, they may not already exist
  • Everyone on the project (client and agency) should know what the goals are

Working processes are unclear

So often that it hurts, a contract between client and agency would be agreed, and then the discussion would start on how it was going to be achieved. Typically, agencies wanted a brief, and then wanted to go away for a time, and come back with the answer, and be paid. What we wanted was to collaborate. Have some of their people on site, work over there sometimes, have regular catch-ups to see how it was going. Entirely different assumptions were made about how it was going to be done, and neither the commissioning manager nor the contract team had any idea that this was an issue for the contract.


  • Agree what the workflow will be between client and agency, and include it in the contract or as an appendix
  • Involve the people who will be doing the work in shaping the agreement

Agencies don’t think they need help from the client

This one frustrates me hugely. Almost without fail, the agency pitch in response to a client brief would mostly include platitudes about the business, their customers and processes that were already well known. That’s ok to the extent that it shows that the agency understand something, but it doesn’t help when it’s presented as the big reveal. Then, much of the rest of the pitch is stuff that’s already been thought of but simply won’t work for very good reasons, or is actually a bad idea. After that, there may be a nugget or two to latch on to, which the agency is told to go away and develop. It’s a massive waste of everyone’s time.

Many times I and colleagues have spoken to agencies as they worked on a presentation for the big bosses, offering our help. Talk to us, we said. Share what you’re thinking, and we can help you to take out the road bumps. We’ll tell you if we spot issues with your proposals, if there are political issues you may need to be aware of etc. We made it clear, truthfully, that we weren’t going to steal their thunder, and also that they didn’t have to take out advice. If I had been agency-side I’d have jumped at the offer. Sadly, I can’t think of a single agency that did, with predictable results.


  • Clients should offer expert help in shaping agency proposals
  • Agencies should ask for it
  • Work collaboratively, leveraging the expertise of each organisation


Tech comes and goes. How do we know what will stick? Don’t ask a journalist.

How do you know if new technology is going to stick? It’s a tough question. Sometimes it might be obvious – but then obvious turns out to be wrong. It might depend on who thinks it’s obvious. There’s something of an idea that the good futurologists tend to be right either about the tech, or about the timescale, but not both, although I’m not sure what it means just to be right about a timescale.

Too many years ago I was helping people at work to understand how computers and PCs worked. At the time, the main sizes of available hard drives were 10mb and 20mb. That’s megabytes, not gigabytes. I recall that a manufacturer was talking about drives with hundreds of mb, and I read tech journalists saying that it was pointless. What would anyone use all that storage for? There was the same sentiment as Intel incrementally brought out the 286, 386, 486 etc processors. Why do we need faster computers when my word processor works fine?

I wrote the business case for ba.com (British Airways’ website). It included the words to the effect of ‘There is a risk that the Internet is just a short-term fad’. My joke is that we’re still finding out whether that’s true or not. Around that time I had a 1-1 with a director of the company after he’d had a liquid lunch, and he was strongly of the opinion that no one would ever want to do their grocery shopping online. I suspect that he never actually did any grocery shopping, and it was his wife who did.

Did you know that Microsoft created a spec for tablet computers in the early 2000s, and some manufacturers produced them?

Tablet PC from the early 2000s

They ran a tablet version of Windows XP. I desperately wanted one, but never did get one. They were big and heavy, and I recall seeing Steve Ballmer trying to convince a sceptical journalist that Windows was a suitable operating system for such a device. It wasn’t, and whilst I believe the machine had  niche market, it never did what the iPad did. Here was a good idea, possibly before its time, but also one not executed properly. I don’t think that Microsoft could conceive of a different OS, or of modifying Windows sufficiently to make it appropriate for the device.

Many devices are conceived before their time, and fail because the tech isn’t up to it. The idea becomes discredited, so it’s harder to implement when new tech arrives. The promise of  1-1 marketing and extreme personalisation has been around for more than 20 years, possibly longer. Someone was always working on it, but the technology just didn’t have the grunt and sophistication to make it work. Now the likes of Facebook and Google have it so refined that many people think that those companies are listening in on their computer microphones, and offering ads based on eavesdropping. They aren’t, but Arthur C Clarke said that any technology sufficiently advanced is indistinguishable from magic, and as people tend not to believe in magic these days, they have to rationalise what they see in terms of technology that they do understand. So the mic is the obvious choice.

3D TV didn’t take off. It’s not a bad idea, but the tech isn’t up to it. It doesn’t fulfill a need, or even a want, in a way that’s easy and works for people. I suspect that in one way or another, the idea will come round again, whether it’s VR, or holography, or just doing it better.

Some technologies aren’t going to work. At least, not in the way they are marketed. WAP stands for ‘Wireless Application Protocol.’ In the 2000s the phone companies got really excited, and advertising was everywhere announcing that the Internet had arrived on your phone. I had a Nokia 7110 WAP phone.

Nokia 7110

It was awful. And the experience of the Internet as a clunky dot matrix mainly text experience was worse. It discredited the idea of mobile web for many people who believed the hype and invested in a new phone.

I’ve already given an example above of journalists who didn’t believe advancing a technology. Part of the problem, it seems to me, is that tech journalists aren’t any better than anyone else at imagining how a product that they aren’t interested in could be of use to someone else. If they don’t like an idea, chances are that neither do their mates because they’re all tech journalists as well, and they all reinforce their own prejudices. They also often lack a historical perspective. When the iPad first came out, I took one look at it and it was obvious to me that someone would start making keyboards for it. But none of the journalism I read had a clue about the corporate environment, where I could see it being so useful. It was acclaimed as a ‘media consumption’ device that may or may not survive. It didn’t take long for the keyboards to appear.

I happen to be quite a fan of the Guardian’s tech podcast. Yet a couple of years ago they would regularly dismiss the idea of drone delivery. Amazon announced that they were going to experiment with it, and the podcast team dismissed it as a publicity stunt. There were too many issues, they said. But I reckoned that if planes full of people could fly pilotless, then drone control wasn’t going to be an insurmountable challenge. Drones are being used in Africa now for regular delivery of medical supplies, and the likes of Amazon continue their research.

Similarly with driverless cars. I’ve heard many tech journalists question why we would want them, and whether people are ever going to feel safe in them, and what if the software had to decide between killing the occupant or the cyclist. These are all things that will need to be addressed, and they will be, but they are not harder than other problems that humans have overcome. When the railways came along, there was a fear that humans would be damaged by being exposed to the wind caused by the ‘extreme’ speed of the trains. Even if the occupant of a driverless car is killed instead of the cyclist, there will still eventually be fewer fatalities, serious injuries, and minor scrapes.

What we’re hearing about all time at the moment, and I do mean all the time, are robotics, AI, and VR/AR. Sometimes there’s a confusion of terminology. Robots have been around for decades doing menial repetitive tasks. The sci-fi concept of a robot is of a humanoid device, with human intelligence, so that wraps in AI. Much software these days is touted as ‘AI’, devaluing the term. AI should at least be something that learns, but how that’s different from an algorithm that populates a database with individual peoples’ preferences is a philosophical matter. Whilst there have been great strides with the likes of IBMs Deep Blue, and Alphabet’s DeepMind, there’s still nothing with the general intelligence of a five year old human. But all these technologies will progress, along with quantum computing, which will have niche applications.

My favoured technology of the moment is voice. A bit old hat, but new in the way it’s now being used in tech. You’ve been able to dictate to a computer for decades, using software that you train to understand your voice. Then along came Siri, and OK Google. Now there’s the Amazon Echo, and Google Home. It’s about what’s appropriate to the context. I don’t see that we can entirely replace touch interfaces with voice (maybe we can replace them with thought interfaces…), as the world would be even more of a cacophony, but it’s about finding the appropriate use.

I’ve written in a different post about how voice interfaces will be particularly beneficial to people with a range of disabilities, including dementia and the aged. I was initially uncertain about voice interfaces until I bought an Amazon Echo for my elderly father, and I now have 3 at home. My dad’s had his for a couple of months, and when the wi-fi in his home went down for the weekend he was lost without it. I use voice in my car to dial, and change radio stations. It’s got further to go, and it will go further.

So what’s the takeout of all this?

It’s quite normal for marketing to over-hype a technology, but that can do damage and cause delay in the longer term, as with WAP and 3D.

It seems to me that a few simple principles need to be met if a technology is going to succeed (and these don’t only apply to tech). It must:-

  • fulfill a need or a want
  • do it in a way that people want to use
  • be affordable

That’s the easy bit. It’s harder to know if a new technology will fulfil those criteria in the future, and I’m not sure any of us know how to find out. Just don’t ask a journalist.

It’s easy to sabotage your own checkout. Here are some tips.

The story

My wife asked me to advise on which desktop computers the small firm she works for should buy. Amongst other sites I looked at was Dell, and I’ve covered some of the problems I had on their site in this post about comparing products. I also experimented with putting items in the cart to see what I could do, but found the process obscure and unhelpful, and gave up.

I did eventually buy a couple of computers from Mesh computers, although I nearly didn’t. Here’s why.

Starting off, I was looking for office machines. Here is the main navigation. Where would you look for office PCs?

Mesh main nav

Yep, that’s right, office PCs are in ‘Gaming PCs’.

Mesh Gaming navigation

Ah well. Maybe some customers think they don’t make business PCs, and go away.

In working through the options, I saved one configuration as a quote. After more work, I wanted to delete the quote. Here’s the error that I got.

Mesh error message

I have no idea what this means, other than it’s not letting me delete a quote. Here is the template that I’ve always used for error messages.

  • Tell the user what’s happened
  • Tell the user why it’s happened
  • Tell them what they can do about it

This message gets a fail.

Eventually I figured out the spec I wanted, and added it to the basket. I wanted to buy two of them. Here’s what you get in the basket.

Mesh basket

That ‘2’ in the ‘Quantity’ box was originally ‘1’. When I changed it to 2, nothing happened. When I then pressed ‘enter’, it took me to the next step in the process, but gave no indication as to whether I was buying 1 or 2, or what the total price was. So I stepped back again to this page. I then tried ‘Edit’, which took me back to the start of the purchase process, as if I wanted to edit the spec. I didn’t want that, so clicked the basket link in the main nav, and instead of taking me back to the page shown above, it took me to that next step that didn’t show what I was buying, and expected me to continue! I wasn’t going to do that.

I thought that surely there must be a way of getting an updated price. Eventually, I spotted the ‘update’ button, in the bottom left of the image below.

Mesh checkout

It’s in the wrong place, amidst a sea of identical blue buttons. It wouldn’t be hard to improve the design of this page to be more effective with better use of positioning and colouring. At the least, the main ‘continue this process’ button should be a different and standout colour, and ‘update’ should be next to the things that you can update. Or just make the page update automagically.

When I clicked the ‘edit’ link,  and went though the process, I noticed conflicting information about the configuration.

Mesh configuration

Although I had specified an Intel Core i5 processor, which is shown in the basket on the right, the Core i3 was selected in the configurator. I have no idea what would happen if I didn’t spot this and just continued. The configurator should be sensitive to context. On the first pass it should have the default values. Having returned from the basket, it should show my chosen values.

Having struggled through all this, I paid with Paypal. Here’s the screen asking me to confirm the purchase.

Mesh Paypal checkout

Firstly, who chose the colour scheme? It’s really hard to read, especially if you’re colour blind, like me.

Secondly, there’s no data. Just the price. There are field labels, but no data. The nervous amongst us would abandon at this point. I took it on trust and continued. Someone hasn’t bothered checking whether this part of the process is working properly.


I know from experience that there can be nooks and crannies of a site that the owners and designers rarely visit. This can happen especially for servicing or post-purchase paths, where a specific set of circumstance needs to be set up to trigger a given display. It can also be that the designers just don’t think of certain paths, or don’t think some paths are very important, and if they are not listening to feedback, they aren’t going to find out there’s an issue.

How could these problems have been identified by Mesh?

The only issue I’ve highlighted here that was an error, that would show up in error logs, is the inability to delete a quote, and the designer might well look at it, and think that the system is working as designed. nevertheless, it pays to check error logs, and address those which are happening frequently, or occur at key points.

An analysis of site metrics and paths might lead someone to think there’s an issue, if they see looping, such as when I was trying to update the price for two machines. Visual session replay can also be useful for this. Session replay is offered by a number of companies.

Many of these type of  problems are best found by talking to users, observing their use, and listening to feedback.

In terms of what you pay attention to, it’s not enough to focus on the ‘green’ path, where the customer finds what they want, and goes straight through checkout. You need to ask, at each stage, ‘what might the customer want to know about, or want to do, here’. That doesn’t mean that you have to cater for each circumstance, but it does mean that you have to think about it, and try to get data as to whether it matters or not. You might decide not to cater for an edge case, but if you have a lot of edge cases that you don’t address, then you still have a lot of annoyed customers.

How hard can companies make it to tell them about stuff?

To reply to my own post title, they can make it quite hard.


I’ve posted previously about Wired UK having pages with infinite scroll, but also a footer with links in. So you’re about poised to click a link, and it disappears for more content. I’ve tried a number of different methods to let them know, and never had a response. It still happens. I don’t even know if they know, or care.


On Linkedin, I might do a job search where I’ll put in a job title, leave the location blank, and hit enter, just to see what comes up. This is what.

Linkedin job search bar

It defaults to ‘United States’. What’s that all about? It’s not hard to spot that my profile says I’m in the UK. What usually happens, is that I’ll miss this, start scanning the results, and quickly realise that I don’t know these place names, and remember that they are all in the US. It’s bonkers.

I posted a question on the Linkedin forum, and got a reply from a Linkedin person (I assume), suggesting I use the Linkedin feedback mechanism to ask for a change. Firstly, it’s not something I should have to ask. Someone else pointed out the same thing on the forum months ago, so it’s not new. Secondly, why do I have to tell Linkedin twice? It’s making the customer work to compensate for lack of internal process.

In any case, I previously gave feedback about spam in their discussion Groups, and received no response whatsoever. I don’t even know if it arrived, which hardly encourages me to try again.

There is no sense of engagement with Linkedin as a company on their own social site.

National savings

I have some premium bonds. For those who don’t know, they are effectively a lottery ticket that you keep until you sell them back at face value, run by the government. I don’t have many, but I reckon I’m due a win.

I was looking at my registration card, and noticed that the address is slightly wrong. Instead of e.g. ‘5 Smiths Place, AB1 2BC’, it has ‘5 Smith Place, AB1 2BD’. I tried to fix this online, but can’t log in as my details are wrong! I phoned up. Now, all you have to do is Google the address to find the postcode, and Google the postcode to see it’s not that address. But I was told that I have to write a physical letter before they will change it. Even though they know it’s wrong!


Amazon used to have an email ‘contact us’ button, where you could contact them about anything (in the UK, at least). They now have a locked-down series of question that constrain what you can talk to them about (online) to the topics that they define.

Jeff Bezos, I understand, likes to describe Amazon as ‘the most customer centric company in the world’. Yeah right. He says that ‘helping customers to find the right product for them’ is key to their success. I think that the search on Amazon UK is pretty poor.

Anyway, the relevant point here is that a while ago I gave a one star review to a product, and recommended an alternative. Within a couple of days I had received about 20 ‘not helpful’ votes. I could live with a couple, but this was plainly an attempt to give my review less prominence, gaming the voting system. My reviewer ranking dropped from around 40,000, to 3 million.

If you are going to help customers to find products they are happy with, you don’t want your review system to be gamed. But there was no way online I could contact Amazon about this. I spoke to a colleague who used to work at Amazon, who contacted people there, and the response seemed to be that they couldn’t tell who downvoted anyway, and weren’t really interested.

British Gas

We have a HomeCare contract with BG, which means that we can call them out to fix plumbing problems. We had a constant drip from our cold water cistern, so a guy came round to fix it. The following morning, I went to the airing cupboard to get fresh sheets, and everything was soaking. I tracked it down to a leak from the isolation valve feeding the cistern, in the loft above.

I was due to leave for meetings quite soon, so I was in a mad rush to get in touch with BG to get them to come round and fix it. I couldn’t find a phone number. On the website, you can book an appointment, and there’s a section for ’emergencies’, but that’s only for gas or electrical problems, not plumbing. In the ‘book an appointment’ section, there wasn’t anything that would get someone to me soon.

I tried harder to find a phone number, but couldn’t. So I Googled ‘British Gas phone’, and lo, there was the customer service phone number.

I well understand that companies encourage people to sort out their issues online first, and it makes sense to do so. But there is a but. British Airways went through this cycle of hiding the phone number, which then drove a lot of dissatisfaction. Eventually they realised that you provide the best service and get loyal customers if you make it easy for people to contact you using their preferred method. And yes, work hard at making online easy to use, and a better channel.

With British Gas, I can find their number way easier on Google than on their own site. That doesn’t make sense, and it’s really frustrating.

Eventually, I got through, and eventually, the same guy came round and fixed the leak. I got to my meeting on time.

The lesson

We live in a world of mistrustful millenials and fake news. Whilst companies can’t put unlimited resources on having cosy chats with customers, there does need to be an effort to open communication channels, and to give authentic replies. I’ll maybe post one day about the ludicrous chain conversations you can have with eBay drones.

Companies that want to learn, do better, establish trust, can’t just turn their backs on their customers.

Don’t Start Each Word With A Capital Letter In Headings

It’s something I can’t quite get my head round. Some people have an apparent compulsion to capitalise each word in a heading. For them, it’s automatic, and it’s contagious. It’s also not a good idea. Capitals exist for a reason – they communicate semantic information. A capital letter at the start of a word indicates that it’s the beginning of a sentence, or it’s someone’s name, or a brand name, or a placename etc.

Here’s a link to a BBC news story.

BBC headline

It’s easy to spot that ‘Toby Young’ is a person. If it were capitalised to ‘Row Over Toby Young Education Role’, then you wouldn’t know if it meant ‘Row over Toby young education role’. In other words, it could be talking about ‘young education’, or education for the young. Many times I’ve been confused by headings on websites that fall into this trap.

You’d expect that whether or not to capitalise would be set in standards, but Ryanair doesn’t seem to be sure what to do.

Ryanair headings 1

Ryanair headings 2’Ryanair’ of course, would always be capitalised, but are ‘Gift Vouchers’ a proper noun? Or ‘Help Centre’? Why isn’t it ‘Where We Fly’? At least those sites that do consistently capitalise don’t compromise their brand by looking sloppy. Although, I have to admit that sometimes, even if you have standards in place, you can’t police everyone who puts content on a site.

It’s not just headings that get assaulted by a glut of capitalisation. Product names, and lists of items aren’t safe either. Here’s an example from Boohoo. If they wanted to put in a brand name you wouldn’t easily pick it out.

Boohoo products

I have no idea why Apple would capitalise the options in iOS.

iOS options

Apple don’t capitalise everything on their website, although they do have a strange way of putting a full stop at the end of headings.

Some designs go the whole hog, and capitalise everything. This from Uber, a digital agency.

Uber headings

There can be times when all caps works ok. Here’s a post about it from Saadia M. on Medium.

Finally, here’s an example from British Airways.

British Airways headings

This could seem inconsistent, but we had decided that we would ‘brand’ Manage My Booking. It was a judgement call. We thought that by doing so, and making it a ‘thing’, it would make it a little clearer that it was a destination where you could do stuff, rather than just going straight into a process. You may agree or disagree, but I use it more as an illustration of the thinking. If every heading had been capitalised, it wouldn’t even have been an option.