How to Use AI Writing Software in Your Content Process [Sponsored]

Provided by Writer

Stop! Hopefully, you read yesterday’s post, 6 Tips for Writing Content That Drives an Immediate Response. (If you haven’t had a chance, I recommend taking a look before you continue reading.)

Great! Now we can get to the good part. That article you read was drafted by AI writing technology – specifically, CoWrite by Writer. My objective was to put AI writing to the test, then share a first-hand account of the experience. Below, you’ll find a complete breakdown of every step of my process, from drafting the initial copy to editing and preparing it for final publication.

It starts with a topic and title

I needed to choose a topic for the article to get the ball rolling. At this point, my primary concern was finding the proper focus, so the piece would resonate with the CMI audience and meet the criteria provided to me by CMI’s editorial team. The quality of the resulting draft would be my top concern later in the process, but I just kept it in the back of my mind for now.

After some back-and-forth with the editor, we settled on an idea we both felt would be relevant for the CMI audience: “# Tips for Writing Content That Drives an Immediate Response.

Assuming CoWrite would be able to work with this title and topic, I knew the number of tips included in the article would inform how my first draft would be written. So, I kicked things off by inputting the necessary details into our outline generator.

Step 1. Outlining the article

The setup

My first step was to confirm the appropriate inputs for accurate outline generation. This can vary, depending on which AI writing tools you’re working with; but for CoWrite, specifically, I needed to provide:

• The proposed title (required): 6 tips for writing content that drives an immediate response
• SEO keywords (optional): content, immediate response, writing content
• CTA (optional): left blank

I submitted the inputs and anxiously waited to see if AI could accommodate my needs or whether I’d have to go back to CMI to workshop a different idea.

The AI output

Below is exactly what I received back:

• The Importance of Driving an Immediate Response
• Tip #1: Write a Compelling Headline
• Tip #2: Write to a Specific Person
• Tip #3: Make it Easy to Read
• Tip #4: Use Active Voice
• Tip #5: Use Power Words
• Tip #6: Create a Sense of Urgency
• Conclusion

I recognized it might need some fine-tuning, but at least it confirmed that AI was capable of delivering a draft on the topic I chose. I shared the outline with CMI before moving forward.

My revisions

I used this original outline to create the first draft. Yet later in my process, I decided the AI-generated copy for Tip #5 (Use Power Words) wasn’t the right fit for an audience of experienced marketers. So I returned to the outline generator, reentered the prompts, and requested an expanded set of tips.

One of the additional tips (Add Visual Interest) seemed much more contextually relevant. So, I was able to replace the problematic tip and continue working on the article without having to start all over from scratch.

Step 2: Adding key discussion points  

The setup

The next step was to identify the tips I’d use to support the discussion in each section of the article. No additional inputs were needed here, as I could carry over the tips generated by AI for the initial outline.

At this point, I did take note of the time (2:15 pm), so I could gauge how long it might take to complete the process from here.

The AI output

CoWrite provided multiple tips I could select and apply to each section or modify as needed. In the image below, you can see the options supplied for Tip #3 and how the interface enables writers to reorder key points or add their own.

My revisions

At this point, I could have taken the opportunity to work in some specific stats, quotes, or talking points of my own. However, I wanted to see what the drafted article would look like with minimal intervention. Knowing I could always revisit this step and generate a new draft, I moved on without adding further input.

In retrospect, it might have been helpful to have CoWrite add specific stats and examples at this stage. Since I knew both would strengthen the final article, it would have saved valuable time and effort to rely on AI rather than having to add those details manually at the end.

Step 3. Creating a draft

The setup

After reviewing the key points, I was ready to create a first draft. Again, there were no new inputs needed at this stage – as part of its workflow for writing an article draft, CoWrite simply leveraged the information supplied in the outline.

The AI output

The AI writing tool automatically generated a draft, along with a quality score and a series of improvement suggestions. As you can see in the screenshot below, the objective feedback I received was as follows:

• Overall score: 85
  • The score reflects the number of suggestions compared to the overall length of the article.
• Suggestions: 38
  • This counts up the number of suggested changes related to punctuation, writing style, clarity, and more. Note that I used Writer’s default style guide here, though the AI can also be configured to work with other style guides.
• Grade level: 9.0
  • This score is based on the Flesch-Kincaid readability formula.

My revisions (objective)

Objectively, I accepted the quality score as proof that AI produced a good foundation. Yet I also felt it necessary to read through the article myself so that I could form a subjective opinion on its quality.

I worked through all the suggestions – most of which were related to style or clarity (per Writer’s style guide). While I did get a laugh when it recommended changing “immediate” to “instant” (“use simple words” is listed right under Tip #3, after all), I couldn’t bring myself to make that change.

To complete the initial editing phase, I accepted the remaining suggestions. I also took note of a few things that stood out:

• All of the section headlines were written in title case. They needed to be changed to sentence case.
• Passive voice was commonly used throughout the article.
• In the bulleted sections, the style guide didn’t like the use of capital letters following a colon (unless the next word was a proper noun).

Based on my time stamps, it took me about 20 minutes to address the suggestions sufficiently to move the article into the next phase.

My revisions (subjective)

After working through the low-hanging fruit – grammatical and stylistic errors – I read the article thoroughly to determine how much rewriting might need to be done.

Here, I focused on percentages – was I 50% of the way there? 70%? 90%?! Yet, I also kept in mind the stated intention of this exercise: to keep the article as close as possible to the AI-generated draft while still meeting everyone’s standards (CMIs, yours, and mine).

My conclusion was that the AI-generated draft got me about 75% of the way to achieving that goal without requiring any fundamental intervention on my part. But I did have a few thoughts about what would help bring the article into better alignment with the editorial guidelines I received from CMI:

• In contrast to many of the CMI articles I reviewed for my reference, the AI-generated draft seemed to lack a clear voice or personality. In retrospect, this isn’t surprising. But, to really make the article my own, I would need to invest much more time manually refining the copy.
• While, on the whole, the draft might have lacked a strong “author” personality, there were still passages where CoWrite varied its writing style and approach to make the content more engaging.
• Some sections contained repetitive phrasing or sentences that didn’t really add anything useful to the conversation. Most of the time, I simply removed those passages, though I used Writer’s ReWrite feature (currently in beta) to simplify or enrich some redundant phrases.
• The most challenging requirement was the need to include specific examples and links to relevant source materials. While CoWrite did provide an example of active vs. passive voice, it just wasn’t the right fit for this article. As noted earlier, I would have saved myself some work if I had better leveraged the “key points” step.
• The tips varied widely in the amount of content supplied and how it was presented. For example, the explanation provided for the third bullet under Tip #3 (“Make it easy to read”) was (ironically) too “short and simple” to be helpful, so I had to expand it to provide better value.

Step 4. Editing and revising the article

After writing and editing both articles (the AI-written one and the one you are currently reading), I sent them to CMI for feedback. As you might expect, both required some minor revisions and restructuring on my part before the editorial team moved it into their process for final editing and production.

But there was still one larger piece of feedback to reconcile: The AI-written article needed more sophistication and advanced recommendations to really benefit the CMI audience.

That feedback prompted me to swap out Tip #5 (as referenced earlier) and do some rewriting to strengthen certain points. It also explains my earlier note acknowledging I could have done more during the outline and key points stages to produce a stronger draft.

It’s worth noting that the latest wave of AI content generation technology provides the ability to train AI based on your content. Using that functionality, I could have provided customized input (sample pieces of content) and received an output that was better aligned with the CMI audience’s needs. I would have explored this option if I had not been up against a deadline.

Reflections on writing an AI-generated article

As someone who has not written an article for a third-party publication in quite some time, CoWrite saved me a lot of time and frustration. The initial process of going from title and topic to actual first draft was incredibly quick and efficient, and I spent zero time staring at a blank piece of paper, wondering what to write.

As expected, the heavier lift came during the editing process after I had the first draft. I tracked it as taking from 2:15 pm to 4:37 pm to manage (with some Slack and snack breaks mixed in). A coworker gave it a second round of edits, which brings my estimate up to about 2.5 hours of editing before sharing that draft with CMI.

Addressing the feedback I received from CMI tacked on an additional 45 minutes of editing and rewriting before I submitted the updated draft. Going from title to submitted draft in under 4 hours is a big win, considering how rusty I am at writing.

There are a few tips that I’d provide anyone looking to get started with AI-generated content:

• Start with a strong topic that you feel confident writing about – with or without AI
• Consider all the elements that need to go into a great article and incorporate them into your process (stats, quotes, etc.)
• Think like an editor when working with AI writing, and you’ll get great results.

Now it’s time for the real question: My dear reader, what did you think of the article you read prior to this one? What were your initial thoughts, and what are your thoughts now, having read all the details on how it came together? Do share them in the comments!