ChatGPT, Claude, Bing, And Bard on Social & Email

This article is the third in my three-part AI Showdown series, a competition in which I put the top generative AI chatbots—ChatGPT, Claude, Bing Chat, and Bard—to the test for a content marketing use case.

In the first piece, I introduced the AI Showdown contenders, covering how I use them, what I see as their weaknesses, and how you might use each tool for content marketing purposes.

In the second piece, the showdown began. I walked through the process of asking ChatGPT, Claude, Bing, and Bard to write a blog post on a content marketing topic.

Hint: I was surprised at the outcome!

In this article, the final in the series, I share the rest of the showdown, in which I have the AI chatbots create social media posts and an email designed to lead readers to click in and read the blog post from the AI Showdown, Part 2.

Let’s dive right in.

Writing social media content to promote a blog post

In this part of the experiment, I asked the chatbots to create social media posts.

I didn’t want them to lose sight of the instructions from earlier conversations, so I prefaced my request for the posts with this text:

Can you now write the social media posts to drive people to this blog post? I want one post each for LinkedIn, Twitter, Facebook, and Instagram. Recommend images and incorporate hashtags and emojis where appropriate.

Let’s see how each chatbot performed.

ChatGPT’s social posts

At first, ChatGPT provided me with CSS and other forms of code for the social media posts.

That’s not what I wanted, so I asked it to rewrite without code.

Analyzing ChatGPT’s social posts

In reviewing ChatGPT’s outputs, here’s what I found:

• The hashtags #innovation and #leadership don’t fit the content.
• ChatGPT should know that we can use more than one emoji per post.
• ChatGPT also overpromises in the social posts. For instance, it refers to 10 ingenious ideas, but I saw no ingenuity.
• The word leverage is too formal and stuffy for social media content; I use it in my writing only when discussing exerting force with a lever—or the TV show.

Overall, I gave ChatGPT a passing grade—perhaps a B-minus. It’s not an “A” or possibly even a “B.” But I could work with the content, editing it and enriching it to make it better.

Now, let’s see how Claude’s social content turned out.

Claude’s social posts

Claude jumped right in and provided its social content in seconds.

Analyzing Claude’s social media posts

Perhaps Claude finished so fast because the content was sparse. Here are my other findings:

• Claude seems to have used Twitter’s character count for all social platforms, as the posts are short.
• Emoji usage is sparse; hashtags are few.
• Claude pulls out the we voice whereas the blog post uses the I voice.
• All Claude’s graphics suggestions involve light bulbs; I expected more creativity here.

Claude disappointed me. The model is supposed to be more business-savvy than the others but failed to live up to its reputation.

Next up—Bing Chat.

Bing Chat’s social posts

Bing Chat blew me away with its blog post. Would it do the same with its social posts? I was eager to find out.

Analyzing Bing’s social media posts

I was immediately excited by the size of Bing’s response; it took three screenshots to show it all to you. Other findings:

• Except for its Instagram post, Bing didn’t use emojis. It also didn’t use hashtags.
• The length of Bing’s posts seems right on target, with LinkedIn and Facebook allowing for more content and Twitter/X requiring just 280 characters.
• With warm, passionate phrases like, I’d love to hear from you, Bing’s social posts feel more human than the other chatbots’ posts. Such passion also makes Bing’s posts feel like an actual content creator who cares about their work wrote them, whereas the other AIs’ social posts feel as if a time-strapped assistant who’s just doing their job did.
• I love the Twitter graphic suggestion for a meme of a man looking at a blank paper with the caption, Me trying to come up with new content ideas. That image will resonate with content creators.

Overall, I was surprised by and happy with Bing’s output. Coming into this experiment, I considered Bing to be third on the usefulness hierarchy, with ChatGPT on top, followed by Claude and then Bing and Bard. But Bing provided the best blog post draft, and it won again with its social posts. Would Bing take home the crown? You’ll soon find out.

First, though, look at Bard’s results.

Bard’s social media posts

Bard has been an underperformer in every way and every experiment. Let’s see if it can create solid social media content.

Analyzing Bard’s social media posts

Bard hit one high note with its social content because it was the only chatbot to suggest headlines for its social posts. And those headlines are good because they speak to the benefits readers want—leads and sales.

Here are my other thoughts about Bard’s performance:

• Bard used just one emoji—a lightbulb. And it was for Twitter. In my feeds, emojis are more prevalent on LinkedIn, Instagram, and Facebook.
• For Twitter, the body content is poor, as is the suggestion to add a short video or GIF as the image. More details, please!
• The Facebook body copy leaves much to be desired. I’m a content marketer—the target audience for the content—and would not click to read content promising basic tips on how to research industry trends and the like.
• Bard did not create a carousel post for Instagram—a missed opportunity.
• The body content is too short overall, especially for LinkedIn, Facebook, and Instagram, which allow far more copy than Twitter.

Overall, Bard did what it’s always done for me—fell flat.

Which gen-AI chatbot won the social media post showdown?

And the winner of the social media part of this competition is… Bing Chat.

Although I expected ChatGPT to walk away with the social media content trophy, Bing Chat surprised me by coming out on top, primarily due to its ability to write with passion, making it sound more human—or at least more like this human.

Which tools followed?

1. Bing Chat, first place
2. ChatGPT, second place
3. Claude, third place
4. Bard, fourth place

Now, let’s see how the chatbots performed with the final element of the AI Showdown—an email to drive traffic to the blog post.

Writing an email to promote blog content

As with the social posts, when I asked the chatbots to write the email, I gave them the same instructions:

Great! Now, can you write the text of the email? I’ll send it to my tribe. The goal of the email is to drive people to the blog post.

Let’s look at their outputs.

ChatGPT

ChatGPT’s email came in at 194 words, a good length for my list of busy executives and business owners.

Analyzing ChatGPT’s email

ChatGPT somewhat redeemed itself in my eyes because all my feedback was positive:

• I like the curiosity-piquing subject line.
• The first two paragraphs are what I call “warm-up” text. I could remove it without harming the message.
• “The cursor blinking back mockingly”—a friendly show of attitude that makes the email feel as if it came from a human.
• ChatGPT used bullet points to tease what’s in the blog post, a good tactic.
• ChatGPT also created an engaging sign-off: “To crafting magical content experiences”; I would leave that in if I were sending the email.

However, the overall email is meh—average. There’s nothing offensive about it, but nothing great, either.

Let’s see how Claude did.

Claude

Claude’s email was also a short-enough length for busy readers.

Analyzing Claude’s email

Claude’s email left me holding a mixed bag.

• The subject line—Unlock Content Ideation—puzzles me. Why ideation? Who speaks that way?
• I like the salutation—Hey friends. It’s pleasant and personal but would be even more so if it were in the intimate singular—Hey, friend.
• Claude gets to the point in the first sentence: I just published a new blog post I think you’ll love. (I wouldn’t be so bold as to say love though; like would do it for me.)
• I dislike the essential pathways language, which Claude also used in the blog post.
• Claude did a great job of creatively highlighting unique reasons why the reader might want to read the post—to develop thoughtful evergreen content, uncover inspiration through curation, or inject creativity into your editorial calendar.
• I also like how Claude wraps the email by asking the reader to let me know what they think and offering to brainstorm content ideas.
• I don’t like chat soon as a close, though, because it’s unlikely that I’ll be chatting with more than a handful of people from my list.

Despite several negatives, Claude’s email still comes out on top over ChatGPT’s. I prefer its brevity, creativity, and tone.

Let’s see how Bing Chat performed.

Bing Chat

After Claude’s email, I wondered if Bing might be about to lose the lead, having won both the blogging and social media elements of the AI Showdown.

Also, because Bing’s email is double the word count of ChatGPT’s and Claude’s, I wondered whether the extra words would be worth it.

Analyzing Bing’s email

Bing’s email has pros and cons:

• Pro: I like the subject line; it’s clear and concise.
• Pro: The writing style is friendly and conversational.
• Con: Bing misspoke in saying readers learn how to do the bulleted items; they don’t.
• Pro: The bulleted items will let readers know whether to click in.
• Con: I’m never one to end a list with And much more, exclamation!
• Con: I also don’t like the What are you waiting for? language.
• Pro: I like Bing’s closing, To your content success.
• Pro: Bing invites readers to respond to the blog post and to book an exploratory call.

To answer my question from earlier, yes. The extra words from Bing were worth it. Even so, I still considered Claude in the lead because its email was shorter and more personable.

Now, let’s turn our attention to the final contestant in the final competition of the three-part AI Showdown series.

Bard

Bard’s email was the lightest, coming in at 149 words. Were those words useful?

Analyzing Bard’s email

No, I did not deem Bard’s words useful. Here’s the rest of my feedback:

• Bard pulled in the boost leads and sales language from the social posts, which is good; perhaps I’d tweak the subject line so those words came first.
• I don’t like how Bard simply listed all 10 topics as bullet points; it’d be better to list three or four and provide more detail for each because I crave specificity. I find it’s often better to share more details about fewer things than fewer details about more things.
• The email doesn’t do a good job of selling the blog post, which makes sense because Bard’s blog post was not worth selling.

Bard, not surprisingly, lived up to my expectations of coming in last place.

Which generative AI chatbot won the email-writing challenge?

Although Bing and ChatGPT were closer competitors than they were in the blog-post and social-post writing challenges, neither pulled ahead of Claude. Claude wins first place this time, for all the reasons I outlined.

1. Claude, first place
2. Bing, second place
3. ChatGPT, third place
4. Bard, fourth place

But which AI chatbot won overall? There were four contestants, each performing three tasks. Let’s take a look.

Summary: The winners and losers

In this three-part series, I asked ChatGPT, Claude, Bing Chat, and Bard to write a blog post, along with social media content and an email to drive readers to that post.

To determine the overall results, I assigned points to each placement. The winner of an individual challenge would receive 4 points. Second place would receive 3 points; third, 2 points; and fourth, 1 point. Then I tabulated the results, which follow.

Based on the total points, the overall winner was Bing Chat, with 11 points. ChatGPT and Claude both tied for 2nd place, with 8 points. And Bard brought up the rear with 3 points.

The results were a surprise, as I hypothesized that ChatGPT would win, followed by Claude and then Bing Chat. I was confident that Bard would come in last place, which it did.

The takeaways: How to use each generative AI tool in your content creation workflow

Based on my experiences, including the experience of this AI Showdown, here’s how you can best use generative AI chatbots for content marketing purposes.

Use ChatGPT to:

• Generate initial drafts and raw content you can refine.
• Provide outlines and topic suggestions to spark creativity.
• Answer research questions to help inform content.
• Summarize or expand on information.
• Transform information into different formats.

Use Claude to:

• Review and edit content drafts to improve accuracy, clarity, and flow.
• Check facts and citations to ensure content is correct.
• Give alternative wording suggestions to improve tone and style.
• Evaluate whether the content meets goals and guidelines.

Use Bing Chat to:

• Research the web to find data, examples, and quotes to incorporate.
• Aggregate information from different sources.
• Carry on basic conversations and answer questions for content creators.

Use Bard to:

• Tap Google’s knowledge graph for insightful research.
• Generate fresh perspectives and angles based on broader connections.
• Identify related topics and questions to cover for audiences.

The key is combining each system’s strengths to enhance different parts of the content creation process, from generating ideas to drafting and refining.

And never forget—your human creativity, critical thinking, and oversight are still essential for quality results.