Claude 3.5 Review: A Marketing Copywriter's 3-Month Exploration

In June 2024, I was introduced to Claude. I was about to write it off as noise when a colleague described how they used Claude to rewrite content more effectively and efficiently. Not long after, I wrote a somewhat glowing review, only for Claude 3.5 to be announced literally 5 days later, rendering said review obsolete.

Before I could even think about updating that review though, Anthropic (the people behind Claude) dropped Artifacts. This was a real game-changer for me, and made me rethink how I use GenAI. It helped that the updates came at just the right time, as my team was going through a massive content creation crunch period, and while I didn’t use Claude to write my stuff, I did use it quite a bit to tweak specific content pieces based on feedback.

So, how did Claude do?

Noticeable difference between Claude 3.0 vs. Claude 3.5

There was an immediate difference in the quality of the output, mainly in relation to how well Claude handled input. Whereas previously, like Gemini and ChatGPT, I’d have to prompt it to change the tone and tweak its formatting, this time I did not have to. In fact, out of all the AIs I've reviewed so far, Claude is the only one where the quality of its output gets visibly better with every release. ChatGPT and Gemini's still takes a few days or weeks to show the results - Claude's was almost immediate.

Furthermore, if I gave Claude actual content I wanted it to emulate or tweak, it’d maintain the tone of voice instead of defaulting to whatever it thinks is most appropriate. This carried over to different queries, so I eventually began using Claude more for longer and longer pieces of content because unlike the other AIs, I could rely on it to not change tone on me without warning.

Artifacts makes tweaking copy easy

When working on long pieces of content, one of the biggest problems for all GenAI chatbots is reading space. The current vertical line flow is great for conversations, but it's not that great when you are tweaking output. Bringing up the output in a window next to the chat box is a pretty simple but brilliant idea - it's easy to read and and see where changes are when you update.

I enjoyed Artifacts when it first appeared, but recent upgrades have underscored just how useful Claude is for writers. With Artifacts, you can upload something like the outline of your new blog post into a new chat, and then prompt Claude to suggest keywords, meta titles and descriptions and other suggestions to improve the blog post. Claude does a pretty good job of answering based on your context.

If you upload a longer text, say the actual blog draft, Claude can offer suggestions on what to tweak. Here is where I find it really shines - if you're asking it to rewrite your draft according to specific rules, it will do that. You can also get Claude to tweak specific sections of the draft, and it will do while sticking to the tone you have already suggested.

Plus, it will always generate new drafts which you can easily navigate, so say I like the Intro and lead in from draft 2 but I prefer the description of the various H2 sections in draft 3, I can easily jump between all three drafts to copy and paste what I like.

Yes I know you can just tell Claude to generate a draft in a mix of the above, but I prefer doing the copy pasting into Word myself, as this also forces me to reread what Claude has written. Sometimes (but extremely rarely these days) the word choices can be questionable, even if the idea is sound.

And honestly, if you cannot take the thirty seconds required to read the output before sending it out to whoever needs to get it, you are probably too immature to use GenAI.

Long chats make your limits run out faster

But, it’s not all a great experience. I am on the Free tier of Claude, and so I tend to run out of limits faster. In the last three weeks, I’ve noticed that I’ve come up to the limits faster and faster than I did before, even if I create a new chat as suggested by Claude because it was getting too long.

The great thing is that when you do come up against a limit, Claude will tell you how many messages you have left and when you can query it again. This allows me to plan my work, especially if I am juggling multiple different non-writing tasks for the day.

To add to that, I've also noticed my generation speed is significantly slower compared to others like ChatGPT, Gemini, and even Perplexity. While it's annoying, I don't really mind though, considering the quality of the output. What I appreciate is that Claude doesn't have any limits on the amount of chats I can create, and that it's almost trivial to search through the chats if there's a specific topic I want to revisit. Did I mention that Claude's contextual memory is really good?

What annoys me though, is when I ask a question, and Claude has already begun to answer, but then all the text disappears and I am told by Anthropic that Claude has reached its limits and I'll have to try again later.

Honestly Anthropic if you are going to do this, I'd rather just have a message that immediately pops up to say Claude is having issues rather than teasing me about being under capacity. You're already displaying half the suggested text, which means you already have the answer, why hide it and pretend it's because of server load and issues that the text disappears?

Conclusion

Compared to both ChatGPT and Gemini, I find Claude to be the most useful for marketing copywriters like me. As you can tell, rather than using Claude as a search engine or even note taker, I'm using it as a rewriter and SEO partner. Even with the rate limits, I still think it's one of the better GenAIs for writers, especially when they have an idea of what to say, but for whatever reason, the words just won't come.

Additionally, Claude's free tier is pretty adequate for marketing copywriters, content creators, and anyone who needs a second eye to rewrite, adjust, and optimise their words. With Artifacts now made available to everyone (and on all apps), it's very easy to identify and adjust content according to your needs.

Just make sure you take the 5 minutes to determine what it is you need before writing your Claude prompt. 

Header image by Kaitlyn Baker on Unsplash.