SurferAI is a new AI content writing tool that is going live as part of SurferSEO‘s Content Editor on 23 May. Basically, it’s generating the whole article for you from scratch (you can feed it with an outline though).

In this post, I’m going to test it on two keywords (with different levels of “difficulty”) to see how it does, and show you how to edit them to make them publishable:

1. product engagement score – easy and straightforward SERPs, mostly glossary type and blogs on “what is + how to improve [KW]”; SV – 90, KD – 2
2. ai for product marketers – more of a ‘thought leadership’ keyword, mixed SERP with little content on the topic, and zero SV.

Let’s start from the “easier” post first:

Product Engagement Score

Let’s look at Ahrefs and the SERPs for this keyword first:

As you can see, it’s an easy KW with a nice Search Volume of 90. It’s a generic metric keyword, and the SERP consist mostly of informational-intent articles answering questions such as “What is Product Engagment Score (PES)?”, how to calculate it, why it matters and how to improve it:

Now let’s create a quick outline based on the top articles in SERPs:

Almost all top-ranking posts in SERPs follow the same structure, and consist of the same 4 sections:

1. What is Product Engagement Score?

2. How to calculate PES?

3. Why is PES important?

4. How to improve PES?

Translating it into an outline:

After literally 5-minutes of “dumpster diving” into the SERPs, we have a detailed outline of this article:

H1: What is Product Engagement Score and how to improve it?

H2: What Is a Product Engagement Score?

H2: How Is PES Calculated?

H3: Step 1. Measure stickiness

H3: Step 2. Measure adoption

H3: Step 3. Measure growth

H3: Why PES Is Important?

H3: 1. Identify expansion revenue opportunities

H3: 2. Uncover product-qualified leads (PQLs)

H3: 3. Minimize customer churn

H3: 4. Improve the product experience

H2: Tips for Increasing Your Product Engagement Score

H3: Use interactive walkthrough to help users reach the AHA moment soon
H3: Drive product stickiness with secondary onboarding
H3: Offer in-app self-service support to help users find answers quickly
H3: Use gamification to boost product engagement
H3: Use micro surveys to collect feedback and act on it

H2: Conclusion

Now, let’s see what SurferAI can do with it!

How to create an article with SurferAI?

As usual, start by typing in the keyword into SurferSEO’s content editor:

Then, when the content editor loads, click the purple button “Generate and article with AI” on the top-right:

After clicking that button, you see a few options:

You can adjust the tone of voice (I will leave the “AUTO – based on the SERP” for this one!)

You can select whether you’d like to add your own outline (and yes, we will totally toggle that!)

and as with any Surfer’s Content Editor, you can change the competitors you want to take into consideration (or not) while generating the article:

Now, let’s click “Create Outline”:

SurferAI’s outline vs. “human-made” outline

SurferAI first suggested its own outline:

Compared to the outline I wrote below, it’s maybe a bit less detailed (no H3s on specific benefits or ways to improve PES) but overall – good and roughly following the same pattern as the articles on the first page in Google. It also has the last interesting heading:

H2: Tools to Operationalize Your Product Engagement Score

Which I’m going to leave there. Let’s also add some of the nuance from my outline to this outline and hit “Let’s write” 🙏

In about 5 minutes, we have a ready-made article!

SurferAI-generated article – first draft review:

The article had a Content Score of 79, and overall looked like it followed the outline – but was quite choke-a-blok full of fluff, repetitions and non-sequiturs:

Look at the unedited version here and compare it to the edited version LIVE: https://userpilot.com/blog/product-engagement-score/

Fun fact: We already had an article on Product Engagement Score (PES) but since it was optimized for PES-score, it didn’t do very well in SERPs for “Product Engagement Score” – will the new article beat it?

It took me 1 h 59 minutes to edit this SurferAI article to make it publishable.

👉 If you’re a masochist, you can watch me live-edit this post for 2 hours on YouTube here  

Why “so long”?

1. I went through the entire article sentence-by-sentence, removing fluff, repetition, elaborating on poorly explained definitions, improving the style (which was often too wordy, formal, and convoluted) – e.g. rewriting long blocks of text into bullet-point lists, adding more context, nuance and examples.

2. I added all the images – sometimes I had to look for them in my drive, sometimes I had to find them (save to my computer, upload back to Surfer) elsewhere online – e.g. in other related Userpilot articles.

3. I had to redo the TL;DR section to improve the fragment about PES score definition and calculation.

4. I had to redo the section on tools – it was actually pretty much missing, and just listed some random tools without explaining how they help improve the PES score.

5. I had to add the FAQ section as it was missing – ok, chatGPT helped me with that 😉

6. I had to add the internal links to relevant Userpilot articles – this was obviously completely missing, so I had to find them.

7. I added CTA banners – (contextually, but used some ready-made ones in Poptin – didn’t create anything new)

8. I had to add the meta description, slug and featured image in WordPress – classic before publishing.

Result? Content score went up from 79 to 87. Let’s wait a few weeks to see how this AI generated article performs in SERPs!

But is < 2 hours long to edit an article like that? 🤔

Considering this is typically how long it takes an in-house Content Editor to write a brief for the freelance writer, and then have to wait a week and (for an article of this length – 3,655 words – would have to pay $325 (with a very detailed brief requiring minimal research!) – the 2 hours seems like a massive time and cost saving.

So how to edit SurferAI articles in general?

1. Quickly Skim-Read the whole post and see if all the paragraphs make sense/something is missing
2. Read the piece in-depth and fix any obvious errors (e.g. non-sequiturs, repetitions + fluff – there’s a lot of that)
3. Add more context to instructions and tool descriptions (that are obvious, incomplete, or not actionable = you would not be able to follow the instructions to get the desired outcome.) You can do it using ChatGPT-4, e.g. a prompt “Give me a step-by-step instruction how to use MOZ to build resource page links  in bullet points”; “Give me a step-by-step instruction how to use Ahrefs to build resource page links in bullet points”
4. Add images illustrating the different steps of the instructions – e.g. screenshots of the Ahrefs process.
5. Add images to other paragraphs as prescribed by Surfer.
6. Add internal links to your relevant content + 1-2 outbound links.
7. Add terms from Surfer.P.S. The thought leadership article will come in the next post because your writer got tired 😉