Artisan To Assembly Line

Or How Personalisation Needs To Have Its Ford Moment

I co-authored this essay with Hamid Sirhan in response to the question: “How should marketing adapt to the era of personalisation?”. We submitted it in February 2016 and were awarded the Silver in the annual AdMap Prize competition for it. It was first published in AdMap magazine in July 2016. Don’t hesitate to give Hamid or myself a shout on Twitter if you’ve got feedback.


The 20th century was an era of generalisation where disparate supply chains and production processes were standardised to achieve the scale and efficiencies required for globalisation. Henry Ford became an icon of the age when he observed that the production of automobiles was inefficient and went on to disrupt the incumbent transport paradigm (the horse) by designing and creating the Model T assembly line.

He famously said “you can have it in any color you want, as long as it is black”. In the era of personalisation, we need our Ford moment: we need an assembly line for personalised communication.

It’s within this tension that our argument lies.


We believe that the era of personalisation is something we’re not even close to realising.

Personalisation, you’re doing it wrong.

We all are. Every one of us working in this industry is failing to deliver personalisation in our communications and in our products and services. This isn’t anything to be ashamed of. It’s impossible to achieve at the moment.

We can come close to it, sort of, and only when someone specifically tells us what they want. That’s why search has been so successful since its inception. Paid search continues to grow at a rapid pace (8.7 per cent YOY growth in 2015) and for a format that’s been around for almost two decades, it might be surprising that it’s still worth over £1bn more than digital display advertising in the UK alone.

Search involves fairly simple transactions. You tell a machine what you want in written language, and then it presents you with pre-packaged messaging that tries to cater to its estimation of what your underlying need is. You might get a bingo on the first try, but more often than not you realise that your search terms weren’t specific enough and try again. And again. And again. Until, at some point, relevancy will occur. As marketers, we try to hack relevancy with tricks of the trade like dynamic keyword insertion, which plays off of the organic benefit that will be derived from the mere-exposure effect.

Relevance is the key word here. This communication is only personal to the extent that we’re able to deliver the user relevant messaging off the back of a commonly expressed need. But it’s not really personal. There is no understanding of the individual, unique context.

The advertiser isn’t able to tell that you bought their product last week and, only a few days later, it’s already broken. They don’t know that you’re now more sensitive to brand recognition than when you were initially looking for the product two weeks ago. The advertiser certainly isn’t able to tell… and so on. See. Complete absence of personal context; ergo, impossible to deliver personalised communication, services or products.

Now, for every other type of interaction where the user isn’t specifically telling you what they want, what hope is there? The 41 per cent YoY growth in ad blocking suggests that we’re failing to achieve relevancy, let alone personalisation in our display advertising, despite over $300m invested in ad tech in Q2 of 2015 alone.

But it’s got my name on it

When Coke launched ‘personalised’ products with ‘Share a Coke’, the idea was to alter Coke’s iconic packaging with some of the most common names, allowing people to buy for themselves or gift to friends. It was neat, quirky and led to a seven per cent uplift in consumption as well as generating fame around the campaign. The initiative was, however, a facsimile of personalisation.

Coke engendered the feeling of personalisation by relying on statistical quirks and the Cocktail Party effect to make an impact on Dave and Sarah but it was highly limited; try finding a Coke with Hamid or Ramzi on it. Apparently, anyone with a less than common name is not entitled to unlock happiness without the help of a rudimentary website and substantial international shipping charges.

However, any business that is brave enough to commit to changes in infrastructure at a global level and take a chance on its core product deserves praise — especially when it results in positive business outcomes. But it’s still not personalisation.


We believe that consistently successful personalisation can only be achieved when the entire process is automated.

Attack the block(chain)

Let’s consider distributed ledger technology; decentralised encrypted journals of every interaction and complete historical context of the chain of events relating to the subject of the ledger (both people and businesses can keep ledgers, and those ledgers are able to interact with one another). For the uninitiated, we highly recommend reading the report recently published by the UK Government that describes what is a very complex concept in accessible language.

Distributed ledger technology can do some wonderful things, but for our purposes, let’s discuss how it becomes a component of our personalisation assembly line.

Imagine being able to securely record a complete and unbroken picture of a customer’s relationship with a business. From the first viewable ad impression they’re exposed to, to the first time they return something because they were being optimistic about their waist size, to the date and exact time they decide to phone you to complain about a bad experience. Down to every tap or click they’ve made on your site, ads and emails, across every device.

All of that customer context, constantly built upon, added to and accessible to inform business decisions. We’ve still got problems though. A big one being that we’re not very good at understanding or acting on data.

Most data lives in a museum

Every client we’ve ever worked with has talked about data. The importance of data. Collecting it. Keeping it safe and secure. Analysing it.

But most businesses are drowning in data. Two and a half quintillion bytes of it are created every day. Frustratingly, we’re only able to extract a tiny percentage of the value of this data and our ongoing answer to this situation is to try to collect more of it.

We’re so bad at using our data, that we still rely on asking people, be it face-to-face or via a web survey, what it is they would like us to do to make their lives better. That’s a recipe for consumer-driven disaster like ‘The Homer’ or as Henry Ford is famously misquoted as saying, “If I had asked people what they wanted, they would have said faster horses.”

On a scale of 1 to 10, where 1 is ‘a bit annoying’ and 10 is ‘set-the-world-on-fire annoying’, how annoying is it that we’re still using claimed preferences and behaviours to underpin many of our business, product development and marketing decisions?

Again, this isn’t anything to be ashamed of. Using data is hard. With enough time, data scientists and creative people to hand, we can do something useful with that data eventually. But it’s likely to be a partial picture and the decisions would involve some guesswork, like on an episode of Wheel of Fortune.

And ‘eventually’ isn’t good enough. The value of data has a finite half life.

Oreo’s blackout comms ‘won the… Superbowl’ in 2013 for a similar reason Old Spice Man won the Internet in 2010; teams of talented people were able to respond creatively to data in a short period of time. In these cases, the half life of data and shared, communal context were intrinsically linked. They feel personal insofar as they reflected the context of a very transient zeitgeist to good effect. But still, this isn’t personalisation.

Using data creatively with a short turnaround is award-winning. Using it creatively in real time is divine.

The ghost in the machine

In 2012, a US retailer famously made the headlines for accurately guessing that a young woman was pregnant before even her father knew, simply based on her purchasing behaviour. At a broader scale, we’re able to tap into similar predictive technologies such as Google Now. The next evolution of Google Now, and services like it, will be unlocked by machine learning.

This is far from being science fiction. Google’s AI has already bested one of the top Go players in the world, and MyWave has launched Frank, an intelligent assistant that claims to:

‘protect and respect your privacy. That means you control your data. If you like, Frank can use this information to act as a gatekeeper, filtering out all the things and people that might waste your time’.

So let’s take the concept of Frank, and apply it to the world of communications. Imagine an intelligent assistant; one that you have a deep relationship with. It has full access to every interaction you have with a connected device, be it active or passive, to create an accurate picture of your needs.

Its primary responsibility is to send out a dynamic signal, derived from the data contained within its personal ledger to the web, interacting with vast numbers of business ledgers, informing them about your current context, even down to the level of attention available at any given point in time. Not only will the assistant signal what type of brand, product or service you may be in need of, but also the channels you’re most likely to respond to.

For example, it will signal that product type X needs to be displayed in paid search and within the top three positions for the user to have a high chance of seeing and responding to it. Or it might be a signal that completely reshapes the homepage of your favourite retailer’s website to reduce friction between desire, and fulfilment by populating it with products you actually want, rather than ones you looked at yesterday.

It is the conductor in the world’s biggest orchestra that spans marketing channels, devices and digital experiences; doing its best to ensure the finite resource that is your attention is not wasted. This is another fundamental enabler of true personalisation.

Stock and flow

On the other side of this will be a complex system of ad networks, DMPs, CRM systems and website personalisation engines that ingest the signal from the intelligent assistant and respond to it by delivering the most relevant piece of content available. Currently that content is largely man-made albeit with a smattering of automation. We’ve already covered the use of dynamic keyword insertion to hack attention, and this extends to display advertising, email marketing and website modules that automatically surface products and product categories that you’ve been looking at recently. In the future, this relationship will be reversed to be almost completely automated, powered by human informed constraints and guidelines.

We would argue that programmatic media trading is already being underexploited for the fact that content creation is prohibitively expensive at scale, even when confined to advertising content. If we ever hope to achieve real personalisation and, as an industry, generate serious value from assistive intelligence technology, we will have to find a way through the bottleneck of content creation. What point is there in creating even more sophisticated, one-to-one targeting technology if we can only afford to produce twelve versions of an advert to feed into it?

For us, the only natural conclusion is that this side of the ecosystem also needs to be primarily automated. Creating meaningful content, even Buzzfeed-style lists, is tough but we’re beginning to see AI that can do it. We’ve reached the point where algorithmic writing for certain kinds of styles of content is passing for human-created and credible content businesses like Reuters are using AIs to generate news packages, albeit with a significant degree of original-content input. We’ve seen the automation of music content generation and images too. Even voice and video could be automated within certain contexts.

When combining AI that can create meaningful content with a series of brand levers and controls; distinctive brand assets, topics of conversation etc, you get an endless series of possibilities for truly personalised and meaningful content across multiple platforms served on-the-fly that’s measured and continues to learn, feeding results back into all of the associated ledgers.

At this stage, we think a generalised formula to drive successful personalisation has emerged.

It comes as no surprise to us that there are a very small number of organisations that have skin in the game at every stage of this formula. And If we were the betting type, we’d put our Bitcoin on Google being the first organisation in the world to enable an end-to-end system as we’ve described it.


We believe that by unlocking full automation below the line, we’ll give ourselves a new remit to return to our roots and focus on building brands that last.

Adapting to a new era

It might appear as if our argument leads to the end of marketing as we know it or entirely replaces human creativity; it doesn’t. By automating the recognition, capture and conversion of demand across digital marketing channels, we’ll be creating marketing that is both more effective and efficient than what we’re capable of achieving today as well as freeing out our precious capital, both financial and human, to focus our efforts fully into the craft of brand building.

We’re full of hope and optimism that the talented individuals that are today tasked with the creation of banner advertising, email marketing and paid search copy will be liberated to work on challenges that only humans will ever be able to overcome; those that require creative problem solving abilities and can lead to a new era of hundred year brands.

So how should marketing adapt? Brushing up on Stephen King wouldn’t be a bad start.

We believe that the era of personalisation is something we’re not even close to realising.
We believe that successful, consistent personalisation can only be achieved when the entire process is automated.
We believe that by unlocking full automation below the line, we’ll give ourselves a new remit to return to our roots and focus on building brands that last.

If you made it this far, thanks for reading. If you’d like to chat you can ping me or Hamid a message on Twitter.