The status quo of content automation

How useful are text generators, who uses them and what for?

At face value, content automation, text engines and text robots are every web copywriter’s nightmare: Are we about to be replaced? Will artificial intelligence take our place? Or, seen from a different perspective: Could they be the solution to large-scale content, i.e. great amounts of individual content as seen on many e-commerce or tourism websites? We’ve been keeping an eye on this development for quite some time now and have done a bit of experimenting ourselves. Here are our answers to the most important questions …

What is content automation?

Basically, it’s what it says on the tin – content automation is all about the automatic generation of content. Actually, content automation is even more than the generation of specific text based on structured data and artificial intelligence. It also involves the selection of the right content type and format for every channel as well as content distribution. Content automation constitutes the end of static content, so to speak. In this article, we focus on automated web copy. Consequently, text engines generate automated, data-based web copy. Sounds fascinating, right? Which leads us to the next question … 
Textautomatisierung ist keine Zukunftsmusik mehr

Why is content automation useful?

Or is it?
Content automation is heralded as the technology of the future. Web projects are getting bigger and bigger. Consequently, there’s often no cost-efficient way for the respective copy to be produced by real human beings. Still, there are limits to content generation.

The pros of automated content generation 
  • Content automation increases efficiency.
  • Content automation enhances productivity.
  • Content automation makes it possible to realise large-scale e-commerce projects.
  • Automated texts are unique.

The cons of automated content generation
  • Automated text is relatively easy to spot.
  • Content automation only makes sense for very large amounts of text (appr. 1,000 texts per month).
  • Automated text is not suitable for every type of web copy.
  • Automation tools are very costly.

Where is automatic text generation useful?

As experienced web copywriters, we’d go so far as to say: We can spot a computer-generated text from miles away! This becomes particularly evident when comparing the two text types with each other.

However, it’s a legitimate question: Is this always such a bad thing? Do you really care about a high-quality, entertaining text when buying a pair of shoes in your favourite web shop?

In our opinion, content automation is useful in the following cases:
  • e-commerce and web shops: large amounts of product texts with a similar structure
  • open data projects: texts that are largely based on data and have to be updated frequently (daily or even more frequently) – e.g. weather, news, sports, stock exchange quotations 

In contrast, content automation is not suitable for the following text types:
  • online magazines and blog entries
  • static websites
  • one-off projects

How do text generators work?

It sounds too good to be true: With only a few clicks, text robots generate thousands of individual product descriptions that would take a copywriter ages to do. However, it’s not all that easy.  After all, text generators operate on structured data – data which has to be compiled and managed.

Common automation software extracts data from a database – for instance, product details for a fashion web shop or football league results. Text engines take the semantics and meaning of said data and turn it into meaningful copy. These texts are further based on words and phrases stored in an additional database.

It’s astounding what artificial intelligence is capable of: For instance, the software can identify recurring wins of a football team and will accordingly add superlatives to the respective copy.

Smart web copywriters will have noticed it by now: Automated web copy is only as good as the data in your database!
Retresco erstellt automatisiert Webtexte | © textengine.io
Content automation is more than text spinning!

In times of increasing search engine optimisation demands, the past years saw the emergence of various text spinning tools. It’s easy to mistake these tools for content automation tools. However, all they do is replace and swap individual words in existing texts – which is anything but unique content!

How is automated web copy generated?

Once you have opted for a certain tool (Awantego, Retresco , …), you can start your very special type of web copywriting.

Here’s an overview of your individual steps:
  1. prepare your databases
  2. create sample texts
  3. create web copy prototype
  4. test your content 
  5. generate automated web copy
Step 1
Prepare your databases
As mentioned before, the quality of automated web copy largely depends on the quality of your data. That’s what makes this first step so important in order to receive high-quality content. Your product information needs to be based on clear and structured datasets. “Structured“ means that any information you want your engine to be fed with has to have the same, uniform structure. This is true for your product-related data (product name, price, item number, features, …) as well as for your language-related data (adjectives, phrases, headlines). An example: If you use singular in one text, it has to be used in all the other texts as well. Otherwise, there are bound to be mistakes in your copy. 

You can also use content for certain personas in this initial step.

Our tip: Keep in mind aspects like SEO and define the respective fields in advance!
 
Step 2
Create sample texts
No matter how intelligent your system is, you have to tell it where to start. In our case, the starting point is a sample text created by a copywriter. This text is the basis for all automated articles and has to contain the following:
  • product data variables such as name, colour, material, ingredients, quantities, etc.
  • suitable places for synonyms – this makes your automated copy more individual
  • potential dependencies to underline certain product aspects (example: an attractive travel offer becomes a last-minute bargain)
Step 3
Create your web copy prototype
Your web copy prototype feeds your sample text into your automation tool. Data fields are generated, dependencies are defined, adjectives and headlines are put in the right place. Basically, your rows and columns of data are connected to the respective places in your copy.

In addition, your web text prototype defines conditions for when to use or not to use certain phrases/words. 
Step 4
Test your content
During the test phase, you will find out how well you did in the preceding phases. At this point, you will see whether your data was well structured, whether your sample text is useful and whether your prototype has worked out.

For testing purposes, the database is fed with a small amount of data, which will generate only a few sample texts. This should be enough to see whether your texts are suitable for a large-scale web project or whether you have to go one step back. If the latter is the case, you have to start all over again until the output of your test phase is more satisfactory. Don’t worry, it’s normal to do several test runs!
Step 5
Generate automated web copy
Now comes the easy part: With only a few clicks, you can generate as many texts as there are data rows in your database.
Automatisch generierter Text anhand des Beispiels Fernseher  | © awantego.com

Is content automation really the future of web copywriting?

We clearly say: „Yes and no“. Or rather: “It depends …”

For one, usability is the key to any automation process. What’s more, not every type of web copy is suitable for automated content. On the other hand, many e-commerce projects simply wouldn’t be possible without content engines.

We’re sure that web copywriting will change over the next years – after all, it always has. However, we’re also sure that high-quality content will remain the one defining factor for online success.

By the way: This text was not generated automatically.

Recommended articles