How We’re Approaching AI-Written Content as an Agency: Key Takeaways

Over the last few years, AI use in business has boomed. According to a 2023 research study, 73% of US marketers said their organization has used generative AI tools. And Gartner research estimates that by 2026, more than 80% of enterprises across industries will have used generative AI in their business.

The long and short of it is, if you’re not using AI in marketing your business, you’re likely falling behind.

AI can assist companies in many areas, but one of the most talked-about areas is the use of AI for content generation. AI tools can help brainstorm ideas, give overviews of topics, look for research, write outlines and more – and it does it quickly. Over the last year, the Conveyor team has researched the role of AI in content generation. We have taken a unique approach to AI content generation, ensuring that our clients are bought into its use and that we go over everything with a fine-tooth (human) comb.

As other companies consider AI content generation for the work they’re doing, we’re sharing a few key lessons we’ve learned along the way.

AI always needs to be paired with human knowledge

One of the most important things we learned is that AI-written content should never replace the work of humans. AI is a great writing assistant, without a doubt – but it’s not perfect.

We saw firsthand how it can streamline work and increase efficiency – but it makes mistakes. And it’s limited by the information that it was trained on, and that data could be outdated, biased or simply inaccurate.

The key is this: when you’re using AI to write content, never just copy and paste it for publishing!

Factor in plenty of time to review and “humanize” your content. Check the facts and sources used. Make necessary edits. Add in high-value links. Adjust the tone and diction to fit your company’s voice.

Be sure to add authenticity to the writing by adding supporting elements like new research statistics, personal anecdotes and personal quotes by subject matter experts. AI is assistive, not autonomous – and ultimately, your brand is accountable for any content that AI generates.

Create content beyond blog posts

When you read about AI and content, you might imagine AI quickly generating a plethora of blog posts. And while AI can do that (to an extent, with the caveats above), AI-written content goes beyond blogs. Using the human-paired approach described above, AI can assist with social media posts, emails, newsletters and other promotional materials.

One of our favorite use cases?

AI can take existing content and turn it into new materials in different formats, or optimize it for different platforms. It can quickly proofread content for critical errors like spelling, grammar and even sentence structure.

Marketing teams can also analyze existing content to see how it's performing, asking AI to offer suggestions for changes or improvements.

These use cases scratch the surface of the ways AI can assist with content, and with marketing more broadly. I’m excited to see how the tools continue to learn and evolve.

AI is a starting point for research – not the end-all

Remember how we mentioned that people should check the facts and sources that AI uses? This point cannot be overstated enough!

Sometimes, it’s completely wrong.

And we aren’t the only ones noticing this - 86% of generative AI users are concerned the information that AI delivers is wrong.?

When AI generates inaccurate or misleading answers, it is referred to as an AI hallucination.?

The problem is, It can be hard to spot this erroneous information. And there are consequences if you don’t find it - like the lawyers who used ChatGPT for legal research and submitted at least six bogus cases to the court in a research brief, and had sanctions imposed on them.

Now, not all research done by AI is flawed.

But leaving inaccurate or biased information in your content - even accidentally - can harm your business’s reputation and trustworthiness.

AI can still be a great tool to get research started but here again, it’s critical to include a very careful human review as part of your process.

Train AI to your business voice

You might know that AI models can be trained to generate information. But did you know that your business can also train AI to be tailored towards your brand? When exploring AI for our own agency’s brand, we found that training AI tools made the content results more helpful and applicable to our business needs, and more representative of “who we are” as a brand.

So what does it mean to train AI?

Essentially, this means providing relevant information to help guide the content. Specify the tone AI should use, who the target audience is, what the target audience cares about, what your unique value propositions are, and so on.

Adjust parameters on things like AI’s length of response, number of responses and relevance.

You can also provide feedback in real time and have the model make changes until the work is satisfactory. (Some of our team members have commented that training an AI tool is a little bit like onboarding a new and always-willing-to-please company intern.)

Some businesses are uncomfortable with the idea of using their company data to train AI tools. I get that. But AI companies get it, too, so there are innovative options that allow users to control their data. One example is ChaptGPT Team.?

With this plan, ChatGPT doesn’t train its models on user business data or conversations. Everything lives in a secure environment that is not accessible in the public domain.?

AI and SEO: yes or no?

Does AI help with SEO? It can, but it can’t do it all by itself. AI is not great at things like keyword research, header structure and making pieces flow while incorporating SEO keywords. And you can’t use it to just write 100 articles at a time - Google will always be cracking down on low-quality, unoriginal content created at scale to increase search ranking.

Not all is lost, though. AI can still assist with writing SEO-focused content.

Using AI as a framework for your content is a great way to streamline your content brief and direction. AI might also help you identify new angles. But if you think you can replace your content strategy and writers with AI, you’re missing the mark. Soon, the ecosystem will be flooded with AI-generated content from every industry. You need your human team to help you stand out.

That said, we suggest using AI as the starting point to produce a draft that includes keywords or phrases.

Now, don’t just copy, paste and post.

Read that again: DON’T just copy and paste your AI copy and walk away.

Review and rewrite sections to make sure the piece is in your company’s voice, then integrate authentic elements like quotes or personal anecdotes.

For long-form content or pillar pages, add in helpful graphics or diagrams.

Then check to make sure all the proper SEO strategies are in place.

You’ve still saved some time with the writing while also sticking to your human-designed SEO strategy.

Being transparent with AI use is key for any business

Transparency is important, so consider how you can use AI ethically and responsibly. Be thoughtful and intentional with your AI usage. Following privacy regulations and guidelines is a good start. If you’re accustomed to outsourcing your marketing, make sure your external resources are transparent about AI (and ideally have AI guidelines and standards in place).

Above all, to stay in front of the AI curve, our team has learned that open, ongoing discussions are essential to both our internal team and our customers.

Remember, a human-first strategy will always be the gold standard – with AI being a rockstar sidekick to strengthen and elevate that strategy.

The time you save with AI can be reinvested in working with your team to come up with policies that will guide decisions in this rapidly evolving era.