Testing Article Forge for Creating Content

note: this post includes examples from Version 1 and Version 2

First, I want to mention that I clicked on an ad while reading an article about content marketing to learn about this service. That’s something I usually don’t do, so props to the folks at Article Forge for good marketing efforts.

In early 2020 a new updated version of Article Forge was released, with some of the bugs worked out and improved data compilation. I debated whether I should entirely remove the review of Version 1 and start with a blank page here while I review the changes and updates made in Version 2, but I’m hoping the previous review will help illustrate the the differences.

I’ve added a Table of Contents here so that you can jump to the Version 2 review sections if you’d like…

Second, a quick overview of what Article Forge is. I’ll use their description;

Article Forge uses incredibly sophisticated deep understanding algorithms to automatically write articles in the same way that a human does. These deep understanding algorithms allow Article Forge to research ANY topic, exactly like a human does. Article Forge reads millions of articles, learning everything it needs to know so that it can write about any topic in its own words. This means that Article Forge is the only tool ever released capable of automatically writing high quality articles.

Article Forge is not a typical “content generator” that simply scrapes the web and mashes sentences together. Article Forge writes each sentence in its own words, meaning that Article Forge is able to completely pass Copyscape. This means you never have to worry about Article Forge returning duplicate content. Article Forge also supports Tier 1 and Tier 2 content options with nested spintax that can be used across hundreds of sites.

After generating this content, the program also allows to edit the content and post directly to WordPress sites from inside the dashboard, including creating scheduled posts.

That all sounds great in theory, but those are some pretty big promises to make about something that auto-magically creates an article from information it gathered online. Let’s see how it works out.

Signing Up for a Free Article Forge Trial

There’s a five day free trial. I write pretty much every day, so I figured five days was a good trial period for me to decide if this would be a useful tool to add to my toolbox. Signing up for the trial was easy enough, you enter your name and email address, create a password on the first screen, then on the next screen you add your payment details. Either a credit card or PayPal information are accepted there.

Since you have to add your payment details in the process, please make sure to keep track of the days during your trial and decide in advance of that five day expiration if you want to keep the tool or not. I am admittedly a forgetful person and I know this about myself, so I like to set a reminder on my Google Calendar for the day before a trial expires so that I remember to go back and either cancel or continue, so that was the first thing that I did after I clicked submit. Then I took a look at the dashboard.

It’s worth noting here that the cancellation process is pretty easy. You simply go to your profile, click on the Cancel Info, read the page and type the sentence they ask you to type to confirm that you wish to cancel and then provide any feedback you have about the tool.

Generating Your First Article Inside Article Forge

Once you’re signed up for a trial you can generate your first article right away, so I jumped right in. There are some options to choose from, pictured here at the right.

You start by entering your keyword and any secondary keywords that you want the article to be about.

Then on the content needs slider you choose whether you want “most readability” or “most variations” (also labeled as Money Site Filler, Tier 1, and Tier 2 content).

I’m not a fan of writing for keyword density and it’s certainly not a best practice these days, so I recommend just sticking to most readable. We want content for people to read, not keyword stuffed junk for bots. Of course, that’s just my recommendation. This option is percentage based, so you may want to experiment with different percentages to see what you get back.

Next, you’ll choose the length of article you’d like it to generate. There are four choices; Very short 50 words, Short 200 words, Medium 500 words, and Long 750 words. That’s pretty standard, so not really much to comment on about that.

Then you have some on/off choices to make….

Article titles and headings. I’m going to experiment with both in my review below and see what we get.

Add an image (probability) and Add a video (probability). These are interesting … there’s not really an explanation, but from what I can tell it’s just how likely you are to add an image or video and it will include phrases in the article like, “see the image” or “watch the video below.”

Replace keywords with links. When you turn this feature on it lets you enter the links you want to use for the article, but it seems to place them on the keywords in a random location. I’d rather just add the links later myself.

Guarantee Article Uniqueness. This is a nice feature, if you connect your Copyscape account to Article Forge it will check it automatically. It notes that it may take more than one credit to run this checking process.

Add spinning. Not a fan of spun content, so I probably won’t use this feature. Again, not a best practice these days, but something you can experiment with to see what you get out of it.

My First Article Result was Not Great, Not Even Good (Article Forge Version 1)

Like I said, I jumped right in and created an article right away. I entered the keyword ‘cross stitching‘ and a secondary keyword. I left the slider set all the way to most readable for the content needs. I chose a length of 500 words. I set the image and video probability to 10%. I left the other settings off.

Once I clicked submit the tool worked for awhile. I wasn’t sure how long it would be, so I got up to refill my water, I took out the recycling and checked the mail, had a smoke break, and when I came back it was ready. I’d say less than 10 minutes total.

The article pictured here to the left is exactly what the tool generated without any editing by me. You can click on that image to make it full size for reading if you’d like.

The article generated was 572 words, a decent length. Unfortunately, the content of the article was pretty awful and most of it didn’t even make sense. Starting right in the first paragraph; the second sentence made no sense and was very poorly worded.

And it just got worse from there. It mixed sentences about cross stitch, crochet, embroidery, and beadwork together. The whole thing was just a mess. It described the wrong supplies, and then even for the supplies it talked about the information it gave was incorrect in some cases.

This paragraph in particular was pretty darn funny:

I don’t know why they’re stitching on a pumpkin, but hey whatever makes you happy.

So I deleted this article, but I wasn’t ready to give up yet. I figured cross stitch is a bit of a specialized topic and it’s also a small niche, so maybe the generator would do better if I went with something more general or just in a different niche.

My Second Generated Article Result, Still Not Great (Version 1)

SO let’s try this again. For my second generated article I opened up Niche Reaper and picked a random keyword phrase with a high search volume and a lot of indexed pages in the search results.

The keywords entered for this test were ‘apartments in Indianapolis‘ and secondary was ‘apartments Indianapolis’

The settings were most readability and 500 words, with the article title and headings option toggled on. I was a little more mindful of the time the tool took on this test since I stayed at my desk, working on this review post in one window and watching the Article Forge progress bar in another window. It was about 6 minutes for this one. I will note here that my kids are in the other room playing League of Legends so there is a possibility that my connection was slow today, so your results may be faster.

The article generated for this one was 503 words and still not a very good result. Again, you can click on the image for a full size version that you can read if you’d like.

The very first sentence in the article is about choosing a dog and has nothing to do with an apartment. The rest of that first paragraph doesn’t really make much sense either.

The second paragraph starts off, “In other words, some apartments aren’t well ventilated,” but there’s no context for that statement at all. Then it talks about an advertising tip mixed in with how to find an apartment opening, speaking to two entirely different audiences there.

Here’s an interesting paragraph sample from this one…

My Third Generated Article Result, A Little Bit Better (Version 1)

I still wasn’t prepared to give up yet. For my third test I decide to go with a longer keyword phrase in the hopes that being more specific would bring a better result.

I used a keyword phrase of ‘how to start affiliate marketing‘ and a secondary phrase of ‘learn affiliate marketing’ thinking that surely there is a lot of content out there on the subject for the tool to go mine and come back with something resembling a real article. I went with most readability and 750 words, titles and headings off though.

Still under 10 minutes for the result even though I went with the longer article option and the article generated was 787 words. The result was a little bit better than my first two attempts above, but still not something that I would publish on any of my websites.

There are still several parts that don’t make sense and a handful of inaccuracies.

There’s also mention of a tutorial video being below the article, which of course you would have to go find one or create one. Noting here that I had the image and video probability set at the default 0.

Honestly, I don’t think I could even edit this to make it better, it’s just so all over the place. It would be easier for me to just scrap it and start over from a blank page.

Testing Article Forge for Product Review Content (Version 1)

Since I wasn’t having much luck with the general topic article generation, I decided to give product specific content a try.

First try on product focused content was ‘Ninja blender reviews‘ with the settings at most readable and the Very Short 200 word option. I was thinking maybe it could generate a blurb to use as part of a promotion.

The program generated 210 words in less than 5 minutes. It’s not completely awful, but it’s certainly not ready for publishing either. Although, in this instance, I think a little editing could bring it up to a usable quality.

That’s progress, right?

I wonder if it’s a better keyword selection or if it just generates short content better.

Let’s use the same keyword phrase and see what we get when we request a longer article….

Using the ‘Ninja Blender Reviews’ keyword phrase and settings on most readable, I toggled on the headings and titles and also set the image/video probability bars to 10% each, and chose the longest content available.

It came back pretty quickly again with 723 words of blender themed content. While the result was better than the first two tries above, it’s still not something I would publish. You can click on the image here at the right to read the article if you’re willing to subject yourself to some more of this nonsense with me.

The title generated doesn’t really make sense and after reading the article it doesn’t fit the content either.

The first thing I noticed was that it mentions several other popular blender brands, which would actually be great as part of a review site if it wasn’t so all over the place. The references don’t really make that much sense and the paragraphs seem like they are just tossed together.

The third paragraph….

There are just too many things going on in this paragraph and they’re not even things that are related to each other. The first sentence makes no sense. There’s no flow. The grammar is pretty bad. The whole paragraph is just a hot mess.

The closing paragraph starts off, “If you’re on the lookout for a blender that does that, you will likely should spend more income,” but there is no context for the ‘that’ to which it is referring.

There’s an advanced option in the settings under content needs for shuffling paragraphs, I’m not sure I’d want them any more shuffled than they already are.

Testing Article Forge with More Keywords Added (Version 1)

My next thought was that if I added more secondary keyword phrases it might help the tool to generate a better quality article.

More information, better results. Right?

I started with a main keyword phrase of ‘pressure cooker reviews‘ and added four secondary keywords to test this idea. I kept the ‘most readable’ setting for content needs and selected a 500 word length. I turned off the titles and headings option.

Adding more secondary keywords did not improve the article result.

You can click on the image at the right to read the entire article it generated with no edits from me, but to give you a quick idea of how it turned out here’s the first paragraph of the 554 words it provided:

The second paragraph then talks about rice cookers, finding a good price on a new store, and mentions, “A pressure cooker isn’t enjoy a normal Saucepan,” and then recommends a brand of pressure cookier.

The third paragraph talks about hiring a pressure washer versus doing the pressure washing yourself. An entirely different topic. Even if the topic were pressure washing, half the sentences don’t make any sense.

The last paragraph is talking about pellet stoves and I’m not sure even what the last sentence even means:

I didn’t give up here. I actually ran more tests with different keywords in a variety of niches, different article lengths, and so forth. I couldn’t come up with much worth sharing here as part of the examples, let alone publishing to a website, newsletter, or anywhere else. Some results were better than others, but most were not good.

Possibly the Best Result I Received from Article Forge Version 1

Of all the random keyword tests I ran, this one produced the most readable result.

That said, the full article it generated is not something I would publish as-is, it would still need some pretty heavy editing.

The main keyword phrase was ‘how to quit smoking‘ and I added four secondary keyword phrases of’smoking cessation aids, how to quit smoking, quit smoking, benefits of quitting smoking’ to try to guide the tool towards a focused result.

There were a few usable points in the article and it seems to have the least amount of nonsense.

There are a few contradictory points; it mentions both weight gain and weight loss, for example.

I’m raising the white flag at this point, I surrender.

Reviewing the Article Forge Version 2 Update

When you go to Create a New Article much of the settings and input is the same as before.

Start with your keywords, choose your article length, check some boxes, and click the Create New Article button to start generating your article. After a minute or two you’ll have an article to preview, edit, and download on a screen that looks like this…

The title generated for this one wasn’t great, but that’s pretty easy to fix. I wanted landscaping tips for home gardening, though, so I probably should have added some sub-keywords to get a more specific result. The “tips” aren’t the best, but for automatically generated content this is a much better start than things that came out of Article Forge Version 1. I see some sentences that just feel like filler and unnecessary, but at least I can see some paragraphs to expand on. With some editing effort this could become a somewhat decent article.

Let’s try another one...

Generating an Article about Current Events News in Article Forge Version 2

Since it’s 2020 and the world is weird and chaotic, let’s try something in the news and see what happens. I entered ‘murder hornets’ as the keyword and ‘what are murder hornets’ as a secondary keyword.

Well, that didn’t work out very well. It pulled together an article about butterflies and moths, the use of ‘is’ instead of ‘are’ is sort of sloppy, there’s a subtitle repeated four times, the order doesn’t make sense or flow, and there are inaccuracies. There’s a sentence about great horned owls in there which definitely didn’t belong. This one is a dud all around.

Revisiting the Pressure Cooker Article Example

I thought it would be interesting to revisit the pressure cooker reviews example from my original review (above) to see what Version 2 would generate compared to what Version 1 gave….

While better than what Version 1 had generated, this result reads more like a really long fluffy introduction than an actual article.

I’m thinking maybe I shouldn’t have used ‘reviews’ as part of the keywords entered, so I’m going to try again without it. Let’s try ‘best pressure cookers’ instead and see if we get a more product focused result…

That didn’t go much better. The title is decent, but there are maybe three sentences in this one that are actually about pressure cookers, so this isn’t very useful either.

Generating an Article for ‘How to Cook Pot Roast’

The keywords are there, most of the sentences are readable, but it’s still really not a usable article. I feel like I’d be better off just starting over and writing this one from scratch.

What Do Rugs Have to Do with Getting Organized?

I’m not sure, but there’s a whole paragraph about it in here…

Generating an Article on ‘Positive Thinking’

This is the best result I’ve had so far with Article Forge Version 2

The sentences make sense, there’s some better flow here than in other examples. Still, there’s no meat to this content. It’s more like a fluffy introduction than an actual article, but with some work and expanded points it could be the beginning of something good.

Generating an Article on ‘Parenting Toddlers’

This one is a good start, too. With some heavy editing and expanding, this has the potential to become a decent article.

My Thoughts on Article Forge

Content creation doesn’t come with an easy button.

In all my tests of Version 1, no matter what settings I changed or what sort of keyword phrases I used, nothing came out that was even close to publishable. It was barely even readable in most cases. Because I couldn’t get a decent result with the single article generation option, I didn’t even try out the bulk article generation part of the tool.

In testing Version 2, the results were better in comparison to the original program. Still not content I would publish, but some of my test pieces at least gave a starting point for something to work with. I find artificial intelligence projects and automation fascinating, so I hope they continue to tweak and improve the service.

If you want to try Article Forge for yourself and see if you can generate some useful results, I certainly encourage you to sign up for the free trial and play around with it a bit. I would be very interested to hear about your results if you give it a go and generate some content, so please do come back here and drop me a comment about your experience.