Unless you're a reactionary, as a marketer you probably understand that automation is creating more jobs than it makes redundant.

Automation, for all the scale it enables, requires human mastery of technology, process flows and customer lifecycles.

Increasingly personal communications also entail the creation of more content, to suit each segment and event you identify.

But as much as marketing continues to be a balance of science and art, it seems that automated copywriting software could be about to polarise the world of content.

Read on to find out more, but if you want to improve your own skills (and make sure a robot doesn't take your job) then book yourself onto our Online Copywriting Course.


I wrote an article about automation and the need for a human touch. In it, I mentioned Wordsmith, an automated copywriting product developed by Automated Insights.

My assertion was that for all automated copywriting can achieve (sports summaries, financial reports etc), it will never master analogy, inference or humour.

Well, Automated Insights picked up on the article and challenged me to a duel (actually, a conference call).

What their head of comms had to say was interesting. Wordsmith is, in effect, a complex and multi-branched version of mail merge.

So, much like the pathways set up for email automation, a copywriter begins by filling in a large and (potentially) complex template (with as many branches/as much variation as the marketer wants to add).

This template dictates what the content will say dependent on values drawn from a CSV file of structured data.

Raw data is transformed into automated narratives.

Content at scale

The technology is currently in beta, but is due to be available on demand in early 2016.

It's use is, of course, for writing content at scale, where employing people to do the work is simply too pricey or too slow.

For example:

  • Creating thousands of persuasive product descriptions from standard manufacturer descriptions.
  • Turning workout or lifestyle data into a narrative that's more enjoyable for users to consume.
  • Creating internal business or performance reports from sales data.
  • Writing press-release style articles about financial results or sports that would previously fall outside of resource.

One interesting example was from healthcare. A company called GreatCall has a family tracking services that allows people to keep track of the location of their elderly parents.

Rather than serving this information to users in a formalised and insensitive manner (GPS, frequency of movement, timings etc), GreatCall uses Wordsmith to create a narrative from the detail.

So raw data becomes 'X left the house to go to Y, they visited Z and their movement was below average for the week' (or similar - this is my text). 

wordsmith example

Generative writing 

Before you start to think this is an advert for one piece of technology, let me get to the two most interesting points.

Firstly, at a time when we talk about the changing skillset of marketers, it's clear that broad functional skills and an ability to quickly adapt to new tech are important.

The idea that generative writing will become an important skill changes the concept of copywriting skills.

The skill in what Wordsmith likes to call 'data-driven copywriting' is about being able to add nuance and complexity to the original template. This is more about logic than it is copywriting.

Of course, words are still important, but without that balance between science and art, writers will not be able to get the most out of the platform.

Automated copywriting needs an understanding of spreadsheets and structured data, as much as it needs lyrical verve. 

Quality matters more than ever

So, where does this leave real-life wordsmiths?

Well, there's no doubt you're safe in most areas. Crafting long-form content, campaign copy, content marketing, none of this will ever be automated.

But where your role borders the functional, such as writing reports or covering news, you'd better be sure you're adding value beyond simply structuring data.

Using analytics it's hopefully fairly easy to prove the relationship between time/effort spent writing (or devising) content and the success of that content (however defined). It won't always be the case, but hopefully there'll be a regression - we certainly see this for Econsultancy articles.

If you're just churning out copy in a perfunctory manner, there's a robot with a pen waiting to stab you in the back.

Ben Davis

Published 27 November, 2015 by Ben Davis @ Econsultancy

Ben Davis is Editor at Econsultancy. He lives in Manchester, England. You can contact him at ben.davis@econsultancy.com, follow at @herrhuld or connect via LinkedIn.

1192 more posts from this author

You might be interested in

Comments (11)

Save or Cancel
Parry Malm

Parry Malm, CEO at Phrasee Ltd.

Just to clarify - the service you're describing is not copywriting - it's emotionless editorial writing. So, taking N elements of data and gap-filling into pre-written text elements. This is a very useful service for high-volume media outlets - for example, Bloomberg reporting on financial reports around the world. The value-add of a human writer in this instance is low.

However - note that there is no emotion, idiomatic expressions, or "brand voice" in the output - and these elements comprise huge parts of what is true copy writing... that is, writing to drive customer response, not just inform.

So, I'm clearly biased, as my company automates copy generation for email subject lines. Still, our research (led by Dr Neil Yager, who's one smart dude) has shown that within a defined variant space, machines can indeed produce more effective copy than humans.

over 2 years ago

Ben Davis

Ben Davis, Editor at EconsultancyStaff


There's no reason the content has to be emotionless, though? But yes, a human has to input that 'emotion' at the template level.

Anyway, I'm off to have my oil changed.

over 2 years ago

Parry Malm

Parry Malm, CEO at Phrasee Ltd.

You're right, there's no reason. However, a pure templated approach has inherent limitations. There's some very exciting research being undertaken by both us and numerous academic institutions in the field of "natural language generation" that goes far beyond a rudimentary templated approach... and this is very exciting.

I, for one, don't think automated copywriting will, for example, write a compelling, long-form content piece. But for short, defined language spaces - it is entirely possible and indeed optimal.

over 2 years ago

Ben Davis

Ben Davis, Editor at EconsultancyStaff

Cool. If there are papers out there, I'd love to read about it. Will have a look, but by all means recommend some reading if you know of any journal articles etc.

over 2 years ago


Matt Long, Content Director at Fresh Egg

Robots have been writing financial wire stories for AP, without any human editing, for some time now. There's a story about it on The Verge from January, if you are interested.

The gist of that report is that, so far, the robots are freeing up journalists to take a more strategic approach, rather than having to write stories at volume.

Be interesting to see how this one develops.

over 2 years ago

Richard Tidman

Richard Tidman, Managing Director at Space 66 Ltd

Not sure I agree with the statement ' Crafting long-form content, campaign copy, content marketing, none of this will ever be automated'.

There is no reason why technology can't simulate the degree of sophistication offered by a great copywriter, given enough time.

If autonomous technology can beat a grand chess master, design drive a car safely and with less chance of incident than a human - why should generating original, creative, effective written communications be out of reach?

over 2 years ago

Ben Davis

Ben Davis, Editor at EconsultancyStaff


I'm by no means an authority on the subject. Parry is though, and as he says above, he doesn't think long form is possible.

If it is does become possible, then I guess we stray into the territory of Joaquin Phoenix wearing high-waisted trousers and chatting up his alarm clock. Nobody wants to see that. ;)

over 2 years ago

Parry Malm

Parry Malm, CEO at Phrasee Ltd.

The Turing test is hard. Anyone who claims they can out write long form text better than humans is a full of sh

over 2 years ago

Parry Malm

Parry Malm, CEO at Phrasee Ltd.

^your site doesn't accept emojis!

over 2 years ago

Jack Simpson

Jack Simpson, Writer at Econsultancy, Centaur Marketing

@Richard - You're right that this type of technology will become more sophisticated over time, probably more so than we can currently imagine.

But the examples you mentioned - chess, driving, designing a machine - these involve logical processes that can be programmed mathematically.

Long-form storytelling requires human elements such as empathy, humour, abstract thought, imagination. I don't think a machine will ever truly mimic that.

Source: Terminator.

over 2 years ago


James Kotecki, Head of Communications at Automated Insights

@Parry Hello from Automated Insights! You may be interested in a 2014 Journalism Practice study showing our content was "not necessarily discernible from content written by journalists." We wrote about it here:


about 2 years ago

Save or Cancel

Enjoying this article?

Get more just like this, delivered to your inbox.

Keep up to date with the latest analysis, inspiration and learning from the Econsultancy blog with our free Digital Pulse newsletter. You will receive a hand-picked digest of the latest and greatest articles, as well as snippets of new market data, best practice guides and trends research.