We Analyzed 17,500 Pages Sentiment with NLP. Here’s What We Learned
Do I really have to analyze the sentiment of the pages in my SERP?
Yes, and in this article, I will tell you why.
We need to start with a short introduction and a bunch of statistics that will blow your mind.
With the rise of natural language processing (NLP), new interesting data points popped up on our on-page optimization map—and one of them is page sentiment.
In SEO, the sentiment is an interpretation of positive, neutral, and negative emotions associated with the content.
There are a few different solutions that translate the sentiment of a page into a number. The most well-known are Google and IBM Watson, although Watson seems to be more accurate in this area, especially when it comes to the whole article analysis.
Those solutions analyze a whole page or a paragraph in terms of words, phrases, and their context. Then, they define their sentiment on a scale from -1 to 1.
Positive sentiment is when a topic is being described favorably. They usually have positive words like “great”, “hero”, “outstanding”, etc. The sentiment is considered positive if the value of it oscillates between 0.25 – 1.0.
Neutral sentiment can contain both positive and negative signals, and the resulting value is contained in a neutral score range, which is -0.25 – 0.25.
Negative sentiment implies the usage of detrimental statements in the content. They may be presenting the topic in an unfavorable way or focusing on the downsides and flaws. These pages tend to use words like “hate”, “weak”, “stubborn”, “boring”, “danger”, etc. The negative sentiment contains in the range of -1.0 – -0.25.
We analyzed 17500 pages ranking in the top 10 to understand:
- Does sentiment have any impact on SERPs?
- What’s the distribution of positive, neutral, and negative sentiment in search results?
- Is the sentiment distribution any different for affiliate keywords?
Our analysis is based on 1000’s of keywords researched in Surfer.
Summary of Key Findings from Our Analysis:
- Out of 17500 analyzed pages ranking in the top 10, 0.26% have a neutral sentiment. Opinionated content does much better across the sample.
- 3.83% of SERPs have 3 or fewer positive pages in them, while 57.6% of SERPs have only positive content in the top 10.
- There are only 5.5% negative pages in SERPs for keywords containing “best” or “review”.
Position in SERP vs. Sentiment for All Keywords
Our first task was to establish the percentage of positive, neutral, and negative sentiment in SERPs.
Among all analyzed keywords, the average of pages with positive sentiment is 87,71%.
Neutral pages are the least common which is also an interesting finding, in that their appearance is SERPs is almost non-existing.
Apparently users (and Google) appreciate content that has a clear sentiment about a given topic rather than more neutral content.
0.26% of pages have neutral sentiment, and 12.03% are negative.
It was a bit of a shock to see the little difference between each position. I though we will see some dependencies here.
Lack of correlation between sentiment and position among the sample showed off that the sentiment does not affect placing a result on a specific spot in SERP. Narrowing down the set of keywords to the SERPs that have the strongest correlation is just a bit more insightful, but still does not provide anything meaningful.
What does this mean? There is no ultimate rule on sentiment, you have to treat every single keyword separately and analyze top competitors sentiment, examples in the second part of the article stand for it.
Variance values are similar for each position in SERP. That means Google is not testing heavily on any of the top 10 places in SERP. It is slightly lower for the top 3 but the values are really close to each other.
Since we’re analyzing 1000’s of business keywords, we can assume that people are wanting to read articles that will solve their problem so they can achieve a positive outcome, or they may be looking for articles that will assure them in their beliefs and decisions.
This is critical if you’re working on reviews, “pros and cons” articles and similar content.
Position in SERP vs Sentiment for the affiliate Keywords
Phrases with “Best” and “Review” are very common among affiliates. They review products, create lists of the best available models so that was the easiest way to extract affiliate keywords from the sample.
No surprises here—there are much less negative pages for affiliate keywords. 5.5% on average for a SERP. After all, to monetize your money keywords and get your commission, you need to write something that is positive and encourage users to make a purchase.
That’s why 94.15% of all pages in the top 10 have positive sentiment as they contain many superior words and phrases promoting the products they are talking about to influence the reader into making a purchase and ultimately earning them a commission.
In this case, there are zero neutral pages among 402 keywords we included in this analysis! Make your mind and write with strong sentiment, be sure that weak spots of the products you review do not overlay the whole article.
Position in SERP vs Sentiment for 25% of Keywords with The Strongest Correlation
Just like with any ranking signals, for some SERPs sentiment may and may not be a relevant factor. So we decided to analyze 25% of all keywords with the strongest correlation. This is what we got:
As you can see - negatives land much more frequently on the top and bottom of the first page.
Number of Positive Results in SERPs
By now we can see that the positive sentiment dominates analyzed search results.
So we asked: how many positive results appear in each SERP?
The data tells us that:
- 1008 have 10 positive results.
- 57.6% of keywords have no negative or neutral pages in the top 10.
When you add 300 almost positive and 154 with 8/10 positive results you get 1462 SERPs that are strongly positive.
The number means that 84% of analyzed SERPs are dominated by positive results.
On the other hand, only 67 SERPs have 3 positive pages or less. That gives us 3.83% of all analyzed SERPs.
Can I always positive and get the job done?
Let’s move to real-life examples now to see if that’s the case...
Positive content is on fire, but it is not an ultimate tactic
Looking at the stats above, it’s tempting to make an assumption that writing positive content will have the highest chance to rank at the top for your phrase.
While this can be correct for most phrases (especially for affiliate-dominated niches), this can be highly misleading for less obvious cases.
Let’s take a look at some examples of NLP Sentiment Analysis in Surfer.
“Fear of having braces”—a mixed sentiment
For this SERP, we have exactly 5 positive and 5 negative results, as seen in the graph above.
The fun part is, when you look into the pages ranking for this phrase, you’ll see they all serve the same intent: they are actionable guides to help overcome the fear of having braces.
None of them state that with braces you’re doomed and you’ll lose friends.
The difference between those pages is that:
- 5 pages focus on the positive aspects of having braces.
- 5 pages agitate the negative aspects connected to wearing braces.
So positive content states: hey, you’ll look great, you will be healthier and happier.
Negative content states: that’s the reason why you’re afraid and we understand.
Therefore the latter has many more negative words and phrases with the names of diseases included.
If your keyword has the same balance between positive and negative sentiment, you must decide on one approach to your content or another. Just avoid sitting on the fence as you may never get to the first page with a neutral sentiment.
I guess that you don’t want to be changing your mind when creating content on your chosen topic. To prepare the most accurate optimization guidelines based on your competitors - exclude pages that represent different sentiment than your content.
In the case of mixed SERPs, you must monitor the fluctuations inside them and the chances are, your target keyword will be dominated by one type of a page or another.
“Recommended vaccinations”—a strong correlation
When we look into content ranking for this phrase, we’ll notice that pages on positions from 1 to 6 use many negative words (connected to diseases).
Those rankings between 7 and 10 write about positive outcomes coming from recommended vaccinations.
If you want to rank for a keyword with a similar setting, using negative words may be your go-to strategy.
Artex asbestos testing - negative sentiment in service description
This SERP consists of 2 pages with positive sentiment and 8 with a negative one.
Compared to what we talked about in the previous examples, in this SERP we have a mix of different intentions.
Most of the pages are highly opinionated, calling asbestos a “health hazard in million homes”. They discuss the risks of asbestos and how they influence our well-being. Just look at pages ranking three, four, and five:
Pages ranking 6th and 7th are promoting services. They help families test for asbestos so their content is value-oriented.
I wonder if they could rank higher if they focused even more on the risks on asbestos?
That’s a great test to run if you have a similar case.
“Is coinbase safe”—Community vs. brands
This SERP is particularly interesting as once again we have mixed user intents. The most positive content comes from brands. Those are articles that confirm that Coinbase is safe.
What’s interesting is… The two most negative URLs ranking here are Quora and Reddit threads.
Therefore we have 5 positive articles about Coinbase, 2 rather negative from the community forums, and 3 negatives from other brands.
One of the negative pages comes from Coinbase themselves (position 8). Apparently, they focus on the problem they are solving, not the benefits directly. Maybe they could rank higher with positive content?
Most businesses are more willing to focus on benefits and positive outcomes in their content.
However, based on the examples above we can clearly see that even sales pages can focus on problems instead of solutions and find a good position in the top 10. That’s why for not-so-obvious keywords (like fear of having braces), analyzing sentiment can lead to unexpected discoveries and choosing the right side can be game-changing for your page.
For most keywords, keep in mind and apply those three takeaways:
- 84% of analyzed SERPs are dominated by positive results. People would rather focus on the benefits than the cons of whatever they are looking for. For the decision-making process, they are not that interested in content that brings attention to the flaws.
- For most SERPs, it will be easier to rank a page that uses positive phrases like “best”, “hero”, “friend”, or “dream”, and others that suit the topic you write about.
- Big sample analysis shows some trends but examples twist them quite often. Every SERP is different, analyze your competitors and find out what sticks to the top.