How to optimize your Android app description for Google with natural Language

An ASO specialist can create a superb app description with the proper key phrases, which is nice for customers, however not for Google’s algorithm. To ensure that the algorithm to categorize your app and present it in your opponents’ Related Apps, it is very important use Google Cloud Pure Language. This software will allow you to perceive how Google’s algorithm ranks your description and learn how to make it simpler for promotion.


Table of Contents

On this article, I’ll clarify why you want Google Cloud Pure Language, learn how to use this software accurately, and what to concentrate to when making ready an outline. This text is predicated on a lecture from the Asodesk Academy.

What Google Cloud Pure Language is and why ASO specialists want it

Google Cloud Pure Language (GNL) is an official Google software that helps uncover the construction and which means of a textual content utilizing machine studying. This expertise can be utilized by Google Search and Google Assistant. GNL was designed to course of massive quantities of textual content and categorize it. The software can spotlight an important phrases in a textual content and decide an app’s sentiment (constructive, adverse, or impartial).

You’ll be able to strive the demo model of the software or get the total model. When sending as much as 5,000 requests per thirty days, you need to use the software without spending a dime.

An ASO specialist wants to make use of GNL as its metrics can affect Google Play algorithms. GNL understands the which means of the textual content from the primary sentence. That is particularly essential for Discover installs, because it helps decide the app’s relevance to a selected class. The higher Google Cloud Pure Language can categorize your app, the extra effectively your app may be promoted on Google Play.

Tips on how to use Google Cloud Pure Language for ASO

Earlier than getting began with Google Pure Language, it is very important accumulate a superb semantic core with a number of key phrases. It’s best to insert extra key phrases at the start and finish of the outline. Repeat key phrases 3–5 instances.

When accumulating a semantic core for Google Play, it is advisable depend on options from ASO instruments, widespread sense, Google search options, your opponents’ descriptions, Google Adverts, in addition to key phrases in evaluations and Google photographs. For extra data on learn how to accumulate key phrases for the semantic core and put together textual content metadata, learn this article.

Let’s analyze all of the GNL indicators and see which ones are essential and which aren’t for an ASO specialist.


Entities are distinctive objects and customary nouns: folks, locations, occasions, and artistic endeavors. GNL assigns essential phrases within the textual content to one of many object classes or the class “different”. Entities embrace not solely phrases but in addition phrases.

Entities has a Salience metric that demonstrates the which means of a phrase for the textual content as an entire. Utilizing it, you could find out how strongly the phrase impacts the outline.

For ASO, this indicator shouldn’t be vital. Nevertheless, the key phrases have to be included in our description and have a minimum of a small Salience (from 0.02). In that case, they are going to be seen by Google’s algorithms. Entities at the start and finish of the textual content are extra essential, so their Salience needs to be increased.


Within the Sentiment part, you could find out how constructive or adverse your textual content is. The rating exhibits the positivity (from 0 to 1) or negativity (from -1 to 0) of the sentence. 0 is impartial. Magnitude demonstrates the diploma of emotion. The upper this indicator, the extra intense the feelings this sentence evokes.

Based mostly on my initiatives, I might say that Sentiment doesn’t have an effect on the categorization of the app or its full description.


Syntax will assist younger ASO specialists learn to write full descriptions on Google Play.

The arrows present the dependency of key phrases within the app. You’ll be able to see that Google identifies an important phrases in sentences and subordinates different phrases to them. For instance, it makes the phrases “order taxi” and “order service” with the phrase “order”. Google must accurately establish the subordination between essential search phrases.

If Google doesn’t acknowledge the which means of separate phrases and phrases, it would misunderstand the entire sentence. Due to this fact, it is very important verify the outline and rebuild the syntax in order that the subordination is appropriate and clear.


Google Cloud Pure Language can attribute a textual content to one of many 700 out there classes. Confidence is very essential for ASO specialists because it signifies which class an app will fall into on Google Play. If the software accurately identifies the class, the app will seem in Related together with comparable apps. Due to this fact, your app class should coincide with the classes of your direct opponents.

Confidence for the required classes needs to be 0.85. You don’t must make Confidence any increased. In my expertise, Confidence above 0.85 doesn’t have an effect on search outcomes and rankings. In case your app has a number of classes, their Confidence could also be barely decrease than 0.85.

Don’t attempt to get into the proper class artificially. In my expertise, I’ve seen that synthetic categorization of an app doesn’t result in something good. It is very important work out the semantic core nicely and put together a full description in order that it accommodates related key phrases. Then GNL will decide what classes your app belongs to.

Not like different Google Cloud Pure Language tabs, categorization solely works for English. If it is advisable verify Class Confidence for the language your app is localized into, translate it by way of Google Translate into English. Google Play scans the outline utilizing Google Translate, so it’s essential to make use of this particular software. You can too adapt the English description for one more language should you see it matches the class nicely.

What impacts Class Confidence:

  1. Punctuation marks

Generally, even a slight change in punctuation can have a big effect on the end result. For instance, within the description of a pdf scanner, I didn’t change the sentences in which means however changed all of the commas with an ampersand (“&”). Earlier than this, Confidence was 0.59.

After I modified all of the commas to “&”, Confidence rose to 0.62. It seems that “&” works higher than commas. When you can’t get the Class Confidence you need, strive changing commas with different punctuation marks.

  1. Morphological development

I found that -ing endings have an effect on confidence. For instance, the outline of an app for scanning paperwork with a lot of -ing suffixes had a Confidence fee of 0.59.

I changed the gerundial nouns with verbs, for instance, “crop” as an alternative of “cropping” and “improve” as an alternative of “enhancing”. Altering a number of phrases helped enhance confidence to 0.77.

Easy constructions make your description extra understandable for Google. Due to this fact, it is advisable use extra verbs in sentences and phrases — this has a constructive impact on syntax and categorization.

Necessary factors to recollect about Google Cloud Pure Language

  1. Key phrases which are essential for promotion should have a Salience above zero.
  2. Google Cloud Pure Language analyzes not solely phrases, but in addition phrases and entire sentences, so it’s essential to concentrate to the syntax.
  3. Crucial factor in GNL is choosing the proper app class. To extend Class Confidence, use easier constructions and rearrange the syntax/morphology.


Skip to toolbar