The key to effectively leveraging AI? A structured workflow

Introduction

“It's the workflow, stupid!”

Paid channels are becoming more expensive, less efficient, and harder to scale. Paid social generates only 2% of leads for most companies. At the same time, Big Tech is snatching up most of the margins.

That’s why owned channels and organic search are now more important than ever, which makes high-quality, original content even more of a priority for marketers. In the age of generative AI, that should be no problem. But are things really that easy?

‍

It's been over a year since generative AI took the world by storm, and the marketing world is still not quite sure what to make of it. To some marketers, generative AI is a game-changer; others find that it fails to live up to their expectations. The main downsides include ChatGPT’s lack of factual accuracy, its robotic style, and simply how superficial and generic the content is.

While generative AI may help marketers produce content at a lower cost, it doesn’t address the problem of scalability.

Is it possible to use AI to create content for multiple channels, targeting multiple audiences in different languages, all without compromising brand safety?

Marketers quickly run into bottlenecks in growing their brand, because they’re not seeing the real problem: workflows. If you want to effectively leverage AI, step one is solving the workflow problem. 

‍

The scalability challenge

Content creation is difficult to scale for a wide variety of reasons, including:

1. The marketing team isn’t mature enough. They don’t yet have a clear brand story worked out, and their content lacks a distinctive point of view. They also haven’t yet figured out how to align their team, channels, content, data, and tech stack to achieve the best results. 

2. Lack of creative talent. Many brands simply don’t have the capacity to create high-quality content. They may also be reluctant to work with experts outside their organisation, or don’t know how to organise this properly.
‍
3. Lack of a structured workflow. Technology is only effective if it’s incorporated into a solid workflow. Most brands haven’t yet found a disciplined way to organise their content creation. 

‍

I’d like to emphasise here: when I talk about “workflows,” I mean the series of steps it takes to get from a marketing goal to a ready-to-publish piece of content. 

Say your marketing goal is to grow your brand. It’s crucial that you translate that goal into a content programme.

But how can you do that when content production is actually one of the biggest bottlenecks slowing you down? What would a solid workflow look like – and where does (generative) AI come into play?

It starts with ideation

Step one in the workflow is defining which content you need to create for which channels. AI can play a creative role in this by helping you with content ideation. With the right prompts, you can let ChatGPT generate content ideas, for example. By “briefing” AI tools even further, using first- and third-party data, you can significantly increase the quality of their output.

There are also loads of free tools online that can help you with this. Just remember that these won’t automatically adapt to your context, such as a specific domain or location. You’ll also want to filter out overly generic results.

Step two is the briefing. In content creation, the rule of thumb is: bad briefing = bad output. Unless you properly brief your content creator to begin with, you’ll wind up in an endless loop of revision rounds.

Before you can create a good briefing, it’s important to specifically define your quality standards. Part of that involves creating a detailed “tone of voice” guide. By setting out these rules from the start, you avoid discussions down the road like “This sounds off-brand” or “This won’t appeal to our target audience.”

‍

If you haven’t yet worked out your brand’s tone of voice, you can also use AI to analyse your blog or texts from another brand that you like.

In addition to your brand’s tone of voice, it’s vital to narrow down your most important SEO keywords. Again, you’ll find many AI tools available online for this task, including AnswerThePublic, Frase, and Clearscope.

Whichever tool you use, be sure to adhere to a few proven best practices. Start with a keyword that captures your main topic or niche. For example, if you want to write about content marketing, use that as your primary keyword. Then use AI tools to generate a list of related keywords. 

Content production

The next step in the workflow is content production. For every piece of content you create, aim to leverage the right balance of AI and human talent. Japanese search marketing expert Kenichi Suzuki reports that Google doesn’t require you to state whether your content is AI-generated. It’s entirely up to you. Google does however advise publishers to be cautious about posting AI-generated content without having it proofread by a human. The same goes for content that is machine-translated.

Google’s algorithms are based on human content, which means “natural content” ranks highest. Human intelligence is also a key component to creating content that abides by Google’s E-E-A-T guidelines, which stand for “Experience, Expertise, Authoritativeness and Trustworthiness.” AI, for example, cannot claim “experience.” That’s something only humans can do.

The workflow I’ve been describing so far is mainly focused on creating top-of-the-funnel content, which tends to be longer, more narrative, and more brand-specific. You can rely much more heavily on AI to create your bottom-of-the-funnel content (such as product descriptions, ad and social copy, and CTAs).

In sum, it’s crucial to find the right balance between human and machine when creating content for each stage of your funnel and each type of content. For example, if you’re creating hero content (such as an eBook) you can use AI to create the outline, H2s, H3s, and key messages, but you’ll still need a human writer with subject-matter expertise to conduct deep research, contribute unique insights and possibly even interview additional experts. As you move further down the funnel, tasks like writing landing pages can easily be left almost completely to AI, with a thorough proofread from a human editor.

The final step in the workflow is localisation, which includes translation. If your brand is present in different markets, you must adapt your content to each of those markets to ensure it's effective. The market for machine translation (AI translation) software has grown rapidly in recent years, giving rise to well-known tools like Google Translate, DeepL, and Systran. You can also use ChatGPT to translate texts. While the quality of machine translation continues to improve, it still lacks a key ingredient that only humans can provide: AI is not smart enough to account for cultural contexts and ever-evolving slang. It creates translations that are overly literal and miss out on the subtle cultural nuances that make human speech so special. Also, machine translation doesn’t take local laws into account.

If you want to translate and localise your content the right way, once again, you’ll want to find the right balance of AI and human creativity. It’s perfectly fine to use AI to create the first draft of a translation, but if you want to ensure your finished text is truly localised (including appropriate use of cultural references, slang, and compliance with local laws), then you must always have the final draft prepared by a human translator, working in their own native language.

‍AI alone is not enough ‍

Although some marketers dream of a day when tools like ChatGPT and Google Translate can meet all their content creation needs, the truth is: we’re still a long way from that point. The brands that are winning the content game today are the ones that design effective workflows that incorporate the best of both human and artificial intelligence at every stage. The best way to think of AI is as an enabler – not a substitute for human talent. It is a powerful tool that helps us humans create and localise our content faster and more cost-effectively.