This review explains in detail the many uses of this new AI content detection software!
Today I’m writing about a piece of software that should be in most blogger’s arsenal if you:
- Use AI content,
- Outsource content, and/or
- Publish any content at all.
Yes, this software is something I believe most bloggers should use unless you know you are a rockstar writer all the time.
What does it do?
Originality.ai is billed as an AI content detection software. It scans content you submit and generates a score. The scoring is a sliding scale from 0 to 100. If the AI score is high, that means there’s a strong possibility it’s AI-generated content.
While the creators of this software sell it primarily as AI detection software, I’m finding it more useful as a content quality checker. I explain in detail below under “Most powerful use for me” section.
How accurate is it?
It’s currently in Beta. It’s not 100% accurate. I ran content I knew was NOT AI and it gave it a high AI score. It tends to give out false positives. BUT, IMO, that’s a good thing. That makes it not just an AI detection software but a content quality checker software. The content it scored highly as AI wasn’t terribly great.
Which brings up an important point. Just because an article is not AI does not mean it’s good. You shouldn’t want to just check for AI but you should want to check for quality generally. That’s why I’m so excited about this software since early tests are showing that it is pretty good at determining great, good, mediocre and garbage quality.
Going forward all content will be run through this tool and writers will need to achieve a certain score. I haven’t determined what that score should be yet… I will figure it out by running most of my successful articles through it.
How can it be used?
Originality.ai has many excellent uses. Here they are:
- AI detection from writers: Detect whether content you receive from writers is
- AI detection from AI itself: If you publish AI content, you can use it to see how much editing you must do in order to have the content not be deemed AI.
- Quality filter: Based on my limited testing (a few dozen articles so far) I’m definitely noticing that what I know to be my best content gets really high originality scores. Rote, boring articles, even if not AI-generated, gets lower originality scores. Decent but not stellar scores in between. So far it’s proving to be a decent-quality filter. Because you can onboard a team, you can require your writers to use it.
- Search ranking predictor: I haven’t ran enough tests to know whether this software can predict whether content will but I will be testing this week. What I mean by this is I want to see if my highest-traffic articles consistently receive high originality scores and whether my poorly performing content receives low originality scores. If this is the case after a great deal of testing, this software could be used as a predictor of rankings and traffic. This would be insanely powerful.
Most powerful use for me
I don’t use AI content just because it takes more time to edit it so that it’s publishable than it’s worth it. I don’t use AI content as-is because it’s not great and Google came out saying AI content is against its web guidelines.
Hence, I don’t use Originality.ai for detecting my own AI content.
However, I’m using it not only to detect whether writers submit AI content but also to assess the quality of the content. So far, I’m finding that high originality scores mean the content is good (based on my assessment of it). This means I can require writers (and myself) to only publish content that hits a certain originality score. I’m still trying to figure out what the minimum score should be.
Invite your writers and VAs into your account so they can use it (mandate that they must use it)
You can easily invite team members to your account. They could be writers and/or VAs to run all content through the system before publishing. This is super important for me since my publishing style is to publish a lot of content across multiple sites as passively as possible. This software will serve as a very important part of the process; a quality control if you will.
Does it replace content optimizers such as SurferSEO and other tools?
Maybe. Too early to tell yet. I will be testing like crazy to find out whether Originality.ai on its own can predict content that will rank well in Google in the long run. Testing and fine-tuning my use of this software will take many months since I can’t come to all necessary conclusions until I get ranking results (or lack of rankings results).
One difficult aspect of this software that I haven’t figured out yet is…
I’m not sure how to change/revise/edit articles so that originality scores improve. With tools like Frase and SurferSEO, you follow the suggestions and your scores rise. Originality.ai doesn’t provide any suggestions. We’re left to our own devices. I actually like this because we need to find our own voices and methods for creating quality content. This means the Web won’t be filled with all the same content. In theory, anyway. We’ll see what happens.
Here are some examples which as you’ll see makes this software pretty promising.
1. Here’s the result from one of the best writers I ever hired. She’s since moved on with other projects of her own but her writing is amazing:
2. Here’s an article from another freelancer who is quite good. The article topic is pretty boring (A simple Q&A article). The article is decent. No need to revise it based on the topic and its potential.
3. Next is the result from the highest-traffic article on my biggest site. It’s a combined effort of freelancers and me. In fact, I’ve overhauled this sucker a few times. I can objectively say it’s very good and deserves its top ranking.
4. Here’s a result from one of the best writers I’ve ever hired who is still writing for me. The article is a very boring topic so hitting the high 90s is probably hard. I suspect it also wasn’t her best effort but it’s good enough.
5. Below is the result from the highest traffic article from my second biggest site. While this article performs exceptionally well, I believe it deserves an 82% based on how it reads.
6. On the same site, here’s an outsourced article that gets almost no traffic after a year:
I actually don’t believe the 33% originality article above is AI, but I can safely say it’s not a great article. It might as well be AI, and that’s the point. It had a whopping 36 visitors over the last 30 days for an okay KW.
Initial Correlation Results Between Original % Scores and Content Success (i.e. traffic)
I ran the ten highest search traffic posts for each of my two biggest niche sites. I also ran several low traffic posts through for scores. What I’m looking for is any consistency between “Original” score levels and content success.
|Top 10 traffic posts (the successful articles)||Originality Score (%)|
|5,167 search visits last 30 days||99%|
|4,747 search visits last 30 days||95%|
|4,607 search visits last 30 days||86%|
|3,583 search visits last 30 days||92%|
|3,268 search visits last 30 days||97%|
|2,859 search visits last 30 days||75%|
|2,848 search visits last 30 days||73%|
|2,546 search visits last 30 days||93%|
|2,395 search visits last 30 days||76%|
|2,390 search visits last 30 days||66%|
|Low Traffic Posts|
|358 search visits last 30 days||98%|
|48 search visits last 30 days||61%|
|48 search visits last 30 days||28%|
|54 search visits last 30 days||44%|
|65 search visits last 30 days||73%|
|Top 10 traffic posts (the successful articles)||Originality Score (%)|
|11,619 search visits last 30 days||82%|
|3,466 search visits last 30 days||72%|
|3,279 search visits last 30 days||75%|
|2,541 search visits last 30 days||96%|
|2,171 search visits last 30 days||16%|
|2,104 search visits last 30 days||84%|
|1,841 search visits last 30 days||83%|
|1,819 search visits last 30 days||98%|
|1,798 search visits last 30 days||89%|
|1,783 search visits last 30 days||89%|
|Low Traffic Posts|
|544 search visits last 30 days||63%|
|19 search visits last 30 days||92%|
|85 search visits last 30 days||97%|
|24 search visits last 30 days||83%|
Findings: It seems to me there is a correlation between high original content scores and blog content success. However, there is not correlation between poor post success and original score. In other words, it seems (based on 30 articles across two sites) a high original score in Originality.ai is necessary but not sufficient for high rankings.
Moreover, it seems that best-performing content has a score of 80% originality or higher. There are some outliers of course (including a post with 16% original score) but 80% and higher seems to be the norm for the top ten posts for each site.
PLEASE NOTE: There are many, many factors that go into SEO. It would be naive and irresponsible of me to suggest that high Original scores guarantee top search rankings. That is NOT what I’m suggesting. I’m suggesting that there seems to be a minimum original percent score threshold across my best-performing content. Which means I should definitely only publish content with a score of 80% or higher. That will give it a good chance to rank but it’s not a guarantee.
How much does Originality.ai cost?
Here’s some more good news. It’s $.01 per 100 words. You pay as you go (my fave pricing for software). A 1,000 word article costs you $.10. If you find that this offers good quality control for your blog, that’s a pittance to pay IMO.
3 Important Points to Keep In Mind
- I have not tested enough articles to know whether there’s a strong correlation between high original scores and high traffic… yet. I’ve tested articles I know are good which always score highly. Mediocre articles score lower. Garbage scores dismally (even if not AI). You must do your own testing and probably quite a bit of it to come to conclusions about anything.
- The software is still in Beta. I should have mentioned this before as well.
- Some topics are boring so hitting 90% or higher might be very hard. You will need to test to determine what scores serve as a “good content” score for your site(s).
Overall, Originality.ai is very cool and as it improves it will be more and more reliable. I will be doing a ton of testing because if there is a strong correlation across my sites of high originality scores and high traffic, that is extremely powerful. I suggest you do the same.
There will always be exceptions of course. Some content does well because of images, charts, tables, etc. That means it’s not really conducive to this type of testing.
BTW, how did this Originaility.ai review score?
Check it out. Yup, this software recognizes stellar content, that’s for sure.
Jon runs the place around here. He pontificates about launching and growing online publishing businesses, aka blogs that make a few bucks. His pride and joy is the email newsletter he publishes.
In all seriousness, Jon is the founder and owner of a digital media company that publishes a variety of web properties visited and beloved by millions of readers monthly. Fatstacks is where he shares a glimpse into his digital publishing business.