On what basis are recommendations displayed on VK. VKontakte has released an updated mobile application

“Chief editor of the GetGoodRank blog, web analyst, blogger.
When promoting a page on any social network, it is important to consider that the news feed is updated every second. It is physically impossible to show the user everything he is subscribed to. We offer a list of recommendations from the VKontakte team to improve the ranking of community posts"

In June 2016, the social network VKontakte launched a smart news feed in which posts are ranked based on many factors of user interaction with publications. The algorithm for determining whether a specific publication should be shown is complex. Recommendations from a social network expert and GetGoodRank will help you understand how to make a post visible to most users.

Variety of posts

Users react differently to posts in the news feed. The choice depends on many factors: the purpose of visiting a social network, personal preferences, and the user’s mood at a particular moment in time. Therefore, it is important to create a variety of posts: videos, collections of images, links to publications on the site, useful short notes. The ranking algorithm for VKontakte posts determines the novelty of the post for the user. Posts containing new information are shown higher in the feed. Conversely, if a publication duplicates information from previous posts, it will be shown lower in the feed.

Publication format

There is no single correct indication of which format of publications on social networks is preferable for the audience. The format also depends on the type of information being communicated, the purpose of the publication, and the target audience for which the post is intended.

Images and viral videos get more likes and shares and are the preferred format for social media users.

To select the optimal post format, it is necessary to analyze the reactions of page subscribers to various publications. What collects more social signals: collection images, short notes, checklists, videos, podcasts.

GetGoodRank recommendations: if the page is just launching and you do not yet have the opportunity to assess which format is preferable for the target audience, analyze the publications on the site. So, when planning a launch on the VKontakte social network, analyze the site’s publications and see which posts have collected the most reposts/likes on the social network. Such an analysis will show which publication format will be optimal for a given social network at the initial stage. This does not mean that publications on VKontakte will have to be limited only to these formats. By publishing different posts, you can track user reactions and understand which post formats do not work and which ones earn more social signals.

The VKontakte algorithm shows users more posts in formats with which they interact more often.

The quality of media content affects the ranking of posts

Even interesting and useful publications will not receive the proper response among the target audience if the quality leaves much to be desired.

User Feedback

Comments are a kind of indicator of the target audience's interest in publications and the page as a whole. You should not close this option. After all, it is in the comments that you can find out what users don’t like, what stops them from buying, and why they prefer competitors. Users often name competitors with whom they cooperate and indicate their advantages. No analysis will give you as valuable information as live user responses.

GetGoodRank: comments are a valuable source of data for improving the site, service, topics for creating blog posts, and studying the pain points of potential clients.

Don't be afraid of negative comments. The correct response to negative reviews will show other users that you are a reliable company that values ​​​​its reputation and cares about customers.

The main reason for closing comments is the fear of spam and obscene entries. To combat such a negative phenomenon, VKontakte offers spam comment filter, customizable in community management. The filter instantly identifies keywords and removes unwanted posts from the community wall.

Posting frequency

The frequency of publications is another determining factor for ranking publications. Posts that are infrequent and interesting to the target audience receive higher coverage, while posts about nothing are excluded from user feeds every 5 minutes.

In addition, frequent uninteresting posts reduce the interest and trust of subscribers in the page. This leads to unsubscribes, decreased reach, and fewer interactions with posts, which in turn affects the visibility of more useful and important posts.

Data relevance

The VKontakte algorithm shows above those publications that are directly related to current events in the world. To increase the ranking of posts in the VKontakte feed, VKontakte recommends using hashtags.

Reposts

Use VKontakte post promotion tools

Available promotion tools (for a fee):

• targeted ads - are shown to the user in the left sidebar;
• promotion of posts- a function that is guaranteed to show a publication in the user’s feed marked “Advertising post”;
• advertising posts in communities- a tool that allows you to select interesting communities, prepare an advertising post and place it on sites in a few clicks without tedious negotiations and cost approvals with page administrators;
• special offers in games- the opportunity to get new subscribers and social signals in exchange for bonuses in popular games.

Track community statistics

For business pages, advanced statistics on community posts are available, displaying coverage, number of interactions, number of new and dropped subscribers, and so on. Previously, detailed statistics were available to communities with 10,000 subscribers or more. Now the threshold has been reduced to 5,000 subscribers.

Evgenia Kryukova

We asked representatives of the VKontakte team some burning questions about Prometheus, smart tape and their plans for the future. Answered by Sergey Paranko, director of media ecosystem, and Andrey Zakonov, director of growth and research for the social network. Hurry up to get acquainted with unique information from the original source.

1. How do your algorithms understand that a post is interesting to a specific user? Can you list the factors that most influence the reach of a post? The quality of the content, the response from the audience - this is clear. Anything less obvious?

We analyze the user's activity in the feed and raise higher those posts with which he is more likely to interact: like, write a comment, stick, open a photo, read an article, watch a video. These actions are very different, we predict them separately. Next, all this is taken into account by a general algorithm, which, based on all factors, performs ranking.

2. At your performance on "MEN2“You said that even little things like the weather and the user’s mood affect the reach of a post. How does the tape understand this?

Behind any post there is a huge number of ranking indicators. Each component is a certain value that describes the post itself, its author, the time of day at the time of viewing the recommendations, the speed of the reader’s Internet connection and many other criteria.

If a user visited the news feed or recommendations section while in a deep forest and with a poor Internet connection, we will not show him a bulky video at the beginning of the feed. The algorithm also takes into account the day of the week and time of day. This way, the morning recommendations feed will contain more news and information content, and in the evening - media and entertainment content.

3. Do you have a document that lists all the factors that the algorithm takes into account? Or has he already become so trained that even you don’t know by what parameters he provides coverage for a particular record?

We know the list of factors, but everything is unique for each user-record pair. Therefore, just by looking at a post, it is impossible to say how good it is, it’s subjective. A post about local news from Moscow in St. Petersburg is absolutely irrelevant to me, and you may not be interested in my news feed about machine learning and social networks.

4. Is it possible to artificially increase the reach of a post by purchasing likes and shares on promotion services?

Not worth it. In the long run, this will definitely lead to lower coverage. In addition to the fact that likes and shares do not play a central role in the ranking of a post, such cheats are highly likely to be detected by us.

However, traffic to a post from references on other social networks can indeed improve the organic reach of a post, and there is no penalty for this.

5. Is there an optimal text length for maximum reach in the smart feed? For example, in the announcement of an article – no more than 500 characters. And in a longread created in the editor, there are no more than 10,000?

There is no optimal length apart from the reader's focus. Everything is based on involvement; if you have an informational message, try to write succinctly, in an information style. If you are trying to attract a person to click on a snippet, the same principle is succinct and engaging. Just don’t fall into clickbait, it will negatively affect your reach.

If you're telling a story, captivate the person. All large forms, typically more than 1,000 characters, are best typed in the article editor, breaking up the text with multimedia inserts.

6. Smart Feed favors high-quality videos (over low-quality ones). What is “high-quality video” for VKontakte? What characteristics does it have?

Videos 720–1080 rub.

7. Does VK try to detect fake news? Is the veracity and reliability of the data important, or only the audience's interest in the recording?

On VKontakte there is no pre-moderation of content - as on other platforms, users can freely publish and discuss topics that interest them. At the same time, anyone can report illegal, offensive or inaccurate content in their opinion using the “Complain” button. We consider all complaints without exception. VKontakte has one of the largest moderation services and we respond as quickly as possible.

8. How do you feel about competitions for the most active subscribers and competitions for subscription + repost? They are allowed by the VK rules, but does this somehow affect the coverage of posts (positively or negatively)? Do you recommend using competitions for promotion in VK?

This is a legal way of promotion, but you need to keep in mind that in this way you are attracting an irrelevant audience to your subscribers. This can have a negative side effect - followers who don't engage with your group's main content can have a negative impact on the reach and virality of your posts.

9. Articles created in the editor receive more coverage than articles leading to a third-party site. What should commercial sites and blogs do in such conditions? How can they attract VKontakte users to their site?

A business group on any social network is first and foremost a community. If you treat it exactly like a media asset, you will succeed. Forcing a user to another site, without providing an alternative, in communication can upset a person. As they say: you want to sell me your product, but you do it without respect.

On the VK platform you can communicate with clients/readers, sell goods, receive money transfers and much more. We are developing our ecosystem and making it as convenient as possible for users. And in this regard, it is also beneficial for a business to develop its site within VK. The article editor is another tool for communication and community building.

10. Will wiki articles now receive less coverage than editor-based articles?

It's not the format that decides, but the user's interaction with your content. An interesting story told with a lot of interaction from readers will receive coverage regardless of the presentation format. Another thing is that it is more convenient to hold the audience’s attention with the tools available in the article editor.

Specifically about wiki pages, they do not fall into the “Recommendations” section on mobile devices, they are basically a desktop tool, so yes, their reach is lower than that of other delivery formats.

11. Some users complain that they do not see interesting and necessary posts. Why is this happening? And do you have any thoughts about follow the Facebook path – make sure that users’ feeds contain more personal content rather than commercial content?

We are constantly experimenting with different ranking models, and the main metric is user engagement. You can always send clear signals to the neuron by hiding a couple of posts from unwanted public pages.

12. You say that you cannot purchase bots at all. How then can the community gain its first audience? If you sign up relatives and friends, it turns out that the audience is still not the target one. And if you buy advertising, no one will subscribe to an empty community. How to find a way out?

People come to you for content. Therefore, your community will not be empty if you tell things there that are interesting to the target audience. Advertising for the target audience is an ideal option to start with. This will give you a core audience that will help you grow organically.

13. How do you understand that a penalty has been imposed on the community? How to remove it?

Looking at the statistics, this can be seen in the decreased coverage. Analyze what you have been doing in recent days, which can be regarded by the platform as unwanted actions: clickbait, aggressive audience withdrawal, unoriginal content, too frequent posting. Stop doing this. And it will go away on its own in two days :-)

14. Are there filters based on topic? (For example, “adult content” is downgraded). If yes, what topics have reduction coefficients?

Everything that is not prohibited by the legislation of the Russian Federation and the rules of the site is possible. Everything that is prohibited is prohibited.

15. A case from our practice. We publish unique articles on the blog every day, and they are immediately parsed by several VKontakte public pages. It turns out that they announce this content first, and we announce it 5-10 minutes later. Does this negatively impact the coverage of our posts?

16. That is, the one who first shares the link is considered the original source? If this is the case, we cannot compete with autoparsers, since they do this automatically, but with us manually, we write a unique announcement. What should we do in this case?

I didn’t share the link, but downloaded the content natively. Embeds and links, in principle, do not make you the original source of content. We don’t know what’s on someone else’s video hosting or website, and we don’t parse or analyze this content. That’s why it’s important to upload your materials to the platform’s native submission tools.

17. How will search engines index articles created using the editor? Wiki articles were well found in searches, but what will happen now?

Articles are perfectly indexed by Yandex from the very start, and by Google already too.

18. Public LIVE led by a VKontakte employee? Is it possible to refer to it as an official source of information?

At one's own risk. This is an unofficial community.

How the Prometheus algorithm works

19. Are living people involved in the work of Prometheus? Or is the decision entirely left to the algorithm? How does he learn? Can you tell me in simple words?

The algorithm is automated and independently searches for new authors. The neuron sometimes makes mistakes. We mark each such error by sending a signal to the network about the wrong choice. Based on this feedback, Prometheus learns.

20. We have been posting quality content that resonates with our audience for a long time, but we have never received the Prometheus fire. How can we influence this? Can you give specific recommendations? I want specifics, and not vague advice like “post the content that your audience likes.”

First of all, the Prometheus algorithm reacts to original materials submitted in native VKontakte tools. Native content delivery formats are extremely important because they allow the algorithm to evaluate the uniqueness of the content and its subject matter. The first is important in order to get a mark, the second is to provide the author marked with “fire” with a good boost in coverage. Entries from such authors appear in the “Recommendations” section.

Almost all entries

Good afternoon everyone. In the last article we discussed in detail. Today we will talk about tape management, namely what it is recommended entries, and how the VK recommendations feed is formed.

In previous articles we looked at the news menu and its capabilities. There is another menu item called “Recommendations”. What is it for? Those users who have been around for a long time have many friends, and they are subscribed to various groups and public pages. Every day something new happens in the life of a social network, and everyone and the communities you are part of create and post new information for everyone to see. As a result, so many posts appear in the feed that the user is not able to not only “digest”, but sometimes even read them. It is for this purpose that the site’s creators came up with a special algorithm that calculates the most interesting and relevant news for you.

If we go to this section, not all records in a row will be displayed here, but only those that were selected specifically for you.

How does the VKontakte website offer us interesting news? There is a complex algorithm here that analyzes who your friends and subscribers are, whose pages you view and comment on most often, as well as who you like and whose or in your community. Based on this analysis, your preferences and interests are calculated, and exactly those posts that are most interesting to you are offered. Most often, VK guesses with advice and forwards the most interesting and relevant for you.

It often happens that you subscribe to communities, but then they become uninteresting to you, and you don’t even go there to read the new posts of the group. Gradually, there are more such communities and people on your list, and your feed becomes clogged with unnecessary information. Recommended news helps you save time and find the most interesting information.

• it is not interesting;
• notify about entries;
• complain.

1. How remove recommendations certain friends or communities? If you select the first item, a window with suggested settings will appear in place of the post. Here you can report the post, cancel the action, or select “Do not show community/user news” by clicking on the corresponding button.

If you select this action, posts from a group or user will no longer appear in the recommendations feed. After clicking the button you will see the following message:

If you change your mind, you can still click “Cancel action”. Once the page is closed or refreshed, it will be impossible to cancel the operation!

It may be that you frankly did not like the post. For example, it may contain an insult, a call to aggression or violence, and other reasons. Then you can “snitch” on the post of the social network administration by clicking the “Complain” button in this section. After this, you can select the reason for the complaint from the drop-down list.

2. Now let's look at the second item of the drop-down menu - “ Notify about posts“. If you select it, then all news from the selected groups or users will be displayed in the recommendations feed, and you will also receive notifications if this community or person publishes a new entry.

For example, we set the settings to “Notify about posts” for the community “Nizhny Novgorod” and for the user “Valery Nikiforov”. We will now receive a personalized notification when the selected community and user add a new .

If suddenly you are tired of receiving notifications, then you need to do the following:

• go to the notifications section (the bell icon at the top of the page);
• then select the entry and hover over the arrow to the right of it. Then select the drop-down item “Hide notification”;
• after that, an inscription will appear in place of the notification that the message is hidden. Below it, click on “Do not receive notifications from.” That's it, now the site will not notify you about news from this community or user.

3. Third item of the drop-down menu – "Complain". We have already said that if a post violates current legislation or distributes prohibited content, you can report this to the social network administration in a couple of clicks. To do this, select the item of the same name in the drop-down menu:

You will then be asked to select the reason for the complaint from the list. We select and send.

What happens next is not our business. The administration of the social network itself will assess the harm and danger of the information and decide how to proceed further with the violator. And that's all for today. Now you know how to rid yourself of various information garbage in your feed and watch only what is interesting to you. Don't forget to share the article with your friends and subscribers on social networks.

They received a new design, where there is no longer a side menu, a new section with a recommendations feed, and in the future the company is going to launch a platform for creating longreads within the social network. I spent a day with the new application and talked with company representatives about how everything would work now.

In the web version of the social network, the news feed already had a “Recommendations” section, but its content was not synchronized with the search tab in the application, where the new algorithm works. In addition, the content in the tab is visually grouped: photos and videos are tiled, posts with long texts take up more space and stand out.

For about half of the posts, you can immediately tell how they got into the recommendations - mainly because they were liked or commented on by friends. The other half ends up in an unknown way: the algorithm showed me in the top recommendations an unfunny (although seemingly humorous) video of an unknown girl with 43 views and one like, not even from a friend.

The feed also contains many short text entries, and sometimes there are publications of celebrities with verified profiles (for me these were Ruslan Sokolovsky and Ilya Varlamov). Overall, the recommendations leave the impression of a more convenient tool for consuming content than a news feed or video section. Due to the removal of unnecessary interface elements in this case, the content came forward and became more neat. By clicking on the card, you can go to the post itself - the entire familiar interface and a button for quick subscription will be available there.

VKontakte calls its algorithm “Prometheus” (in ancient Greek myths, it brought fire to people): when a user’s post is included in recommendations, he receives a flame-shaped tag on his page. This way you can understand where the flow of new comments and likes came from.

But it’s not just Prometheus that works in the recommendations feed, looking for something new and supporting little-known bloggers. There are two more types of selected content: the first is personal recommendations built on the user’s social graph, and the second is trends common to VKontakte.

Representatives of VKontakte said that Prometheus can search for users similar to others. If a social network knows that the artist Durand is popular among users, it will use data collected about their interests to search for bloggers similar to him.

At the same time, the company promises that it will fight plagiarists who do not produce their own content, but earn popularity by stealing it from others. The algorithm will search for the original and try to show exactly it, and we are talking about all media - text, pictures and videos, explains Sergei Paranko, who is developing the social network’s media platform: “If someone took someone else’s content and began to climb on it, then we will always check - and whether this content was uploaded by someone earlier.”

Cosmetics

At the beginning of September, the leading designer of VKontakte, Pavel Knyazev, announced an imminent redesign of mobile applications. In the current form of the application, only half of what was promised has been implemented: the likes have become red (they look much juicier and are more pleasant to press), the buttons have become rounder, many icons have been redrawn and, most importantly, the application no longer has a side menu - there are only five buttons at the bottom.

Visually, this is unusual: the side menu sections are now located under the fifth button, and music and video have been hidden there: when listening to music, you will have to return to this tab to switch tracks. The design of the audio player is still the same, and in the video section there are still no recommendations that appeared on the desktop back in 2015 - now all recommendations are only in the search tab.

The news feed is very reminiscent of a combination of Facebook and Instagram: on the one hand, there is a variety of “heavy” content, the output of which is controlled by an algorithm. On the other hand, there are stories and quick access to creating them by swiping to the right. But now VKontakte has not one, but two algorithmic feeds in adjacent tabs: news and recommendations.

It sounds difficult, especially since the mobile version does not imply full synchronization with the desktop - neither in working methods, nor in visual design. Managing Director of VKontakte Andrey Rogozov explains: the web has long faded into the background, mobile platforms are ahead, and user patterns on them are very different.

Platform for longreads

So far, VKontakte has only the old capabilities for creating posts - writing text posts, taking photos, videos and stories with masks. But by the end of 2017, the social network promises to launch a platform for creating longreads that will be “displayed correctly on all devices.”

Paranko explained that the platform for longreads is essentially an editor for long posts with various attachments, which will allow you to publish complex formats within the social network. Previously, VKontakte publics used wiki markup for this, but it will most likely become a thing of the past - it will not be disabled, but it will not be developed. According to the description, the platform will resemble what already exists on Medium, Facebook and Telegram.

Medium has a catalog of interesting articles, Facebook has a tool for richer text layout, and Telegram has a primitive editor for publishing any large materials with support for fast loading via Instant View. VKontakte says it will support the AMP initiative (this standard is used by Google and Twitter) for faster loading of its materials, including those embedded on external sites.

According to the company’s idea, for the user, reading external materials and those published through the platform should be no different - and the “smart” feed will rank such posts higher than regular links, since the content will load faster for the user. According to VKontakte statistics, within a year of the “smart” feed’s operation, 85% of users switched to it.

How correct the display will be on all devices is still a question. A few years ago, as part of a similar initiative for media, large cards appeared on VKontakte for links to media materials and other external sites - however, the social network application for iPad still does not know how to display them correctly.

Built-in messenger

Back in August 2014, representatives of VKontakte, during a special event, announced plans to release a separate messenger for messaging. These plans were repeated many times by media sources until the end of 2016, until the question “When will VKontakte release a messenger” finally became a meme.

Informally, representatives of the social network have long recognized that the need for a separate messenger has disappeared. Now this is enshrined in the company’s official press releases: it began to separately count the audience of the “built-in VKontakte messenger” - according to the latest data, this is 82 million people per month. 97 million people use the social network every day.

• popular interesting content;
• diversity;
• the opportunity to find new sources of information.

The recommendations include new communities, individual posts with interesting information, and all of them are tailored to your interests and needs. Of course, there is also advertising, but even there, all communities, products and services relate only to topics that interest you.

The link to the section is located to the right of the news feed, almost immediately below your name and photo. The first line in this menu is news, and below it are recommendations. By clicking on this inscription, you will go to the section with recommended posts.

Actually, an attentive user will understand at first glance how to get into recommendations on VK. In appearance, it is very similar to your news feed - posts from different publics are placed one below the other. They, just like in the news, are sorted by time of publication. But, unlike news, here you will only see one entry from each community. It is the records that are selected by the program.

By clicking on a post, you will open it in reading mode, and by following the links, you will visit the recommended public pages, you will be able to view other information in them and, if desired, subscribe.

If the news feed displays posts only from those groups to which you are subscribed or from those personal accounts that are our friends, then in the “recommended” feed - for the most part you will find posts from the wall of those who have not yet become your friends or those communities in which you are not yet a member.

The site integrates a complex algorithm that “spys” on you, your friends and subscribers every minute. Any activity is taken into account: viewing information, comments, likes, reposts on the wall or to your public page, forwarding posts in personal messages, etc. This way the program identifies your preferences and interests, and based on this information, the system produces a list of entries that owe you like it.

I would like to note right away that the algorithm does not work on the site for the purpose of spying on you, but specifically to select important information. There is no need to worry about the safety of personal data and the secrecy of correspondence. No information other than commitment to a particular topic is stored by this program.

Yes, it can be customized according to personal preferences, and even a person who is visiting the site for the first time can handle it. There are no complicated menus here; it’s easy to remember how to delete recommendations in VK the first time. A simple menu automatically appears next to each entry. To expand it, hover over the small checkmark to the right of it. The menu will open on its own. All you have to do is choose an action.

• It is not interesting. The post will simply disappear from view. If you immediately click the blue “hide from feed” button, posts from that community will no longer appear in your recommendations. If you change your mind at this point, click “mark action.”
• Save to bookmarks. The post will be saved in a section that you can find via the link in the menu on the left.
• Complain. If you see information that contradicts the norms of the Criminal Code, contains insults or any negativity, you can. Click on the corresponding line, then select one of the items in the window that opens. The administrator will see this message, analyze the data and take appropriate action. Unfortunately, you will not be able to find out anything about whether the offender was punished.
• Notify about posts. This function is similar to a subscription. You will receive notifications that new publications have appeared in the community you like. You won't miss them like you would with a crowded news feed. These notifications come in a “bell” - which is located in the top middle if you open VK on your computer.

When you decide to opt out of these notifications, you can do so as follows:

• go to notifications by clicking on the bell icon at the top of the page;
• find the name of the post from the community you need;
• hover the cursor over the check mark opposite it;
• select “Hide notification”;
• Click on the “Do not receive notifications from” text under the message that the message is hidden.

Settings for each entry and group are saved automatically and remain unchanged until you reconfigure the feed. You can return entries to the list of recommendations by going to filters.

For this:

• click on the funnel icon next to the link to the news feed in the same menu where the recommendations are located;

• find the name of the group;
• click on the empty white circle opposite it so that a checkmark appears in it;
• click "save".

Music recommendations in VK: how this feed works when selecting tracks

This section works on a similar principle, but is located in a different location. To enter it, open your music. Here you will see four tabs: my music, playlists, recommendations and friends' updates. The third tab will contain tracks that, according to the VK robot, are not yet known to you, but will definitely appeal to you. This is a great opportunity to learn about new audio recordings and artists.

This section is located in the right menu; if you don’t have it, then you need to enable it. To do this, go to “settings”.

In the window that opens, select: Configure the display of menu items\ and select the “Music” item - check the box next to this item and click “Save”. Now the “Music” item will be displayed in your left menu.

The algorithm on which music recommendations work takes into account:

• contents of the user's tracklist;
• likes and comments left by the user from communities and on the walls of other people;
• reposts from communities and other personal pages;
• reposts from personal messages and other activity related to music or information to which it is attached.

The more diverse your tracklist and wall, the more diverse music (any audio) will be offered to you in recommendations.

Important! Some users note imperfections in the recommendations, especially for music. In their opinion, they come across tracks that are far from their preferences. The same claims are made by those who regularly read recommended posts. However, such complaints are rare. Sometimes it is not the algorithm that is to blame, but the user himself, who is not aware of the intricacies of the “Recommendations” section in VK, how this feed works and how he selects tracks.

For example, you read an interesting article with beautiful photos, but did not listen to the song that was attached to it. The algorithm took it into account, and tracks similar to it will now appear in recommendations. The VK administration regularly updates and improves the site. It is quite possible that the answer to the question of “what are recommendations” in VK and how this feed works will be different.