How to eliminate old articles so that they grow dramatically in organics: + 104% of traffic in six months

Blog traffic is not just about new articles. If a blog actively lives for more than 2-3 years, content accumulates in it, and these are reserves that can be raised and made to work. We tell how we squeezed out of old articles + 104% of traffic from the search and how much it cost - without attracting paid traffic.

Spoiler: scoring of archived articles, powerful semantics pumping, content updating, technical optimization. No miracles, only systematic work.

Old content - what's the problem

Last summer, PromoPult (then SeoPult) turned 8 years old. At that time there were about 1,000 articles.

Traffic sources: 62% - organic, 26% - direct transitions, the rest - referral traffic, social networks, mailings and other channels.

In August, we began to change the blog: redesign and usability improvement, a new rubricator, new article formats.

And they wondered what to do with previously published articles. There are many, they attract traffic, but there are problems.

Some have lost their relevance in full (articles about Yandex.Catalog and DMOZ nnada?). Others are partially (for example, reviews of services, half of which no longer work). Still others were like nothing, but I wanted to do better.

The saddest thing was the articles that disappointed visitors. They ranked well, the heading responded to the needs of the visitor, but the information in the article was outdated.

About 10 thousand users from Yandex / Google came in a month, and a good half of them fell on irrelevant articles dated 2013-2015. It is clear that no one was particularly late in the blog, and behavioral deteriorated - the failure rate was disproportionately high (80-90%).

It was possible to act differently - for example, to write new articles on popular topics, and old redirects to them, or try to squeeze out old articles from the search with new ones. Or do nothing and just write new articles on traffic topics.

We decided to use our SEO resource sparingly. So, in order not to lose the acquired, but not to drown in updating the content.

Scoring old articles

We conducted scoring of articles - we unloaded statistics of search traffic for the last year from Yandex.Metrica and, by the Pareto principle, allocated 20% (200 pieces) of publications with the highest attendance.

We focused on updating traffic articles. Why:

• the scope of work is not exorbitant and generally finite;
• working with articles that already bring targeted visitors and have a trust;
• It is likely that the proportion of users from search who receive relevant content on the blog will increase and behavioral indicators will improve.

For 6 months of work (from October 2018 to March 2019), we achieved a growth in traffic from the search for actualized articles at 104% compared with the previous period. How we managed to do this, what happened in the end and how much money was spent - on this later.

How to improve articles

Compilation of the semantic core

What did:

• unloaded from Yandex.Webmaster semantics, which were shown on the blog pages over the past 3 months;
• extended semantics with the help of the native parser of association phrases , and then deepened with the help of the parser of search hints ;
• clustered semantics under the Yandex substation — they also used the “native” clusteriser (accuracy in the range from 3 to 7, the region — Russia).

We received about 30 thousand grouped requests. For each request, the clusterizer issued top URLs - they will come in handy more than once.

Then the pages of the TOP-200 assigned clusters. We did it this way: we searched for a page from TOP 200 in the Top URL column of the file with the results of clustering. The cluster in which the URL was found was assigned to the search page.

Often the pages from the TOP 200 were not among the Top URLs. In this case, clusters were assigned based on two criteria:

1. total cluster frequency is above 500-1000,
2. high relevancy of keys to the content of the article.

The keys from the clusters used to optimize meta tags and page content. They entered them naturally and only as needed - without curved formulations, irregular cases and re-spam.

Competitor semantics

Additionally, we analyzed not only the content, but also the semantics of competitors:

• alternately opened the URLs of competitors from the Top URL column and analyzed their texts in the semantic analysis service from Advego (4-7 URLs are enough);
• wrote out information markers - plus-words, which allowed the copywriter to better understand which questions to open.

Further, these plus words were entered into the statement of work along with key phrases from the cluster and comments (what, for whom and how to write). Here is an example of real TK:

Content Optimization and Update

What made in terms of optimization:

• Title headers were registered manually taking into account the occurrence of the most important keys from clusters (they were previously formed automatically in the CMS).
• Prescribed meta tags Description - they were not filled at all.
• Alt attributes were specified for images - they were not there either.
• Found and fixed broken links.
• They added (where it wasn’t) or changed the subtitles of h2-h3, lists, quotes, and other formatting elements — that is, they made the text readable and visually appealing.

As for the depth of the finalization of the content, it depended on a specific article. In terms of the amount of work done, we can conditionally distinguish three groups of articles:

1. Minimal edits. For example, the article about checking the site for PS filters was practically not changed, since it was already in high positions in basic queries and sufficiently covered the topic.
2. Recycling articles by 30-60%. An example is an article about how to speed up site indexing. Its volume did not exceed 4000 characters - many points were missed. The article was expanded more than 2 times, supplemented with specific recommendations, updated screenshots, added a thematic video.
3. New content. With hopelessly outdated articles did not suffer - they made up the TZ and ordered it from scratch. So did, for example, with a selection of 35 cool landing pages . The old selection was made in 2012, and most of the landing pages were either unavailable or not so interesting.

As a percentage, it turned out like this: articles with minimal edits - 44%, with serious modifications - 37%, new articles - 19%.

Perelinkovka

Internal links are needed for both SEO and visitor engagement. But before the blog article almost did not link.

The blog had a lot of content, so searching for articles for linking did not present difficulties. We did not use the services - everything was done manually only from the position of use for visitors.

Links were set right in the text:

And in blocks:

Update publication date

It would be strange to update the article and leave the old date of publication. Therefore, for all updated articles in WordPress, new dates were set a month earlier than the current date. So the updated articles did not interrupt only the published materials in the feed, but signaled to the search engines that they should be reindexed.

Two updated articles in the feed

Redirects

We tried to keep the old URLs. But it did not always work. For example, there was an article “Six Facebook Tricks” dated 2012. We thought, wondered how to rewrite it, because all the "tricks" are outdated.

Analyzed with the help of Google Search Console requests for which there were shows and transitions. It turned out that mostly users find an article on queries related to the removal of the last name from the Facebook profile. Adapted the article to the needs of the audience - they called it “How to remove the last name from Facebook” and described the working removal method (by the way, try the funny method).

It is clear that leaving the URL of the form * / shest-xitrostej-facebook.html is incorrect, because the topic is different now. Therefore, we posted an article on a new URL * / kak-udalit-familiju-iz-facebook.html and put 301 redirects.

In general, it is better to avoid unnecessary redirects. A small discrepancy between the title of the article and its title in the URL is not terrible. The main thing is that the user by URL understands what the article is about.

Native inserts

In each updated article inserted native ad units on the topic. This is the most accurate way to call the reader to the system - it is not as annoying as banners, and allows you to honestly advertise a product without scoffing at the text of a useful article.

Not a single article - how to update the blog as a whole

The increase in search traffic is associated not only with the optimization of articles. Work was done on the blog as a whole. So we improved behavioral indicators - and this indirectly affected attendance.

Usability and layout

The design of the blog has completely changed. On the main made modern cards of records, added involving images and announcements.

Before the redesign, the main blog page looked like this:

After redesign (and later - rebranding):

The layout was also redone: they changed the font, increased its size and line spacing. The width of the text column was about 75 characters - optimal for reading. Added corporate style CSS for subtitles, lists, quotes, as well as a background image under the heading. Made a normal mobile version of the site.

After improving blog usability, the bounce rate fell to an adequate level.

New rubrication - new front of work

Such headings were in the blog before the start of work:

But they have ceased to respond to the concept of service development, since the focus has shifted from SEO technology towards the integrated automation of Internet marketing.

We made thematic headings - for specific areas of Internet marketing (contextual advertising, content, link building and PR, etc.) and our products, added system news and just news.

Next, it was necessary to spread the old articles on new headings and set up redirects. This is a great job, and we didn’t waste time processing the entire thousand articles. New headings moved articles from the TOP-200, which were selected earlier. The rest remained in the old rubrics that are not visible from the main page.

After transferring articles and setting up redirects, traffic is preserved:

This is the statistics of one article: before the change of rubricator, after it and after rebranding (change of domain)

They also optimized the rubrics pages: they registered the Title and Description and added small text descriptions.

Tagging

Tags in the blog make it easier to navigate overlapping articles from different headings and help show recommended articles.

And we had them - 974 tags for 1064 posts. That is, almost every article assigned unique tags, which made them not usable.

We decided that removing tags from old articles is unproductive - a lot of work and implicit profit. Therefore, they simply ceased to produce them in new articles - we actually use about 30 tags.

Result: double organic growth

After starting work on the old articles, there is a steady increase in traffic from searching them:

The total traffic from the search for October-March 2018/2019 exceeded the traffic for April-September 2018 by 104%. Due to the actualization of the content, we received an additional 35177 visits in six months.

Moreover, the deeper processed articles, the higher was the increase in attendance. This is fully consistent with the logic of the process and indicates the correctness of the action.

Worked in plus and behavioral indicators. So, time on the site is growing for all groups of articles. If earlier users did not spend on the site for more than 30-50 seconds, now the average time tends to one and a half minutes.

What about the money?

The calculations presented here are quite conditional - there are still fixed costs for maintaining and promoting a blog. But there are “bare” numbers that can be divided-multiplied and get the cost of reincarnation of old articles.

A total of 65,600 rubles were spent on updating the articles (payment for the work of the optimizer, copywriter and layout designer). Of them:

• 15540 rub. - for articles with minimal edits;
• 21660 rub. - for serious revision;
• 28400 rub. - for new articles.

An additional unit of traffic from the search (for October-March 2018/2019) cost less than 2 rubles (65,600 rubles / 35177 visits).

Judging by the dynamics, traffic will only grow, so its cost (subject to maintaining the relevance of articles) will tend to zero.

It should be noted that, in contrast to new publications, old articles were practically not supported by paid traffic. Organic, only organic.

If you look at the ways of updating, the cost of attracted traffic is as follows:

• minimum changes - 2.7 rubles (15540 rubles / 5760 visits);
• serious revision - 3.7 rubles (21,660 rubles / 5857 visits);
• writing from scratch - 1.2 rubles (28,400 rubles / 23,560 visits).

The cheapest traffic went to articles written from scratch. But this does not mean that other methods are less effective. Writing articles requires serious organizational resources, and there is no guarantee that new articles will “shoot” more than modified ones. Therefore, it is still better to lay eggs in different baskets.

Interesting observations of the behavior of traffic

When you start updating content, you want traffic to grow like this:

In principle, we have something similar happened. But for individual articles, search engines responded ambiguously to updates.

Intermittent bursts of traffic after updating articles

According to 84% of articles, search engines responded positively to the changes. Almost immediately after the update, there was an increase in traffic from organic matter. Even articles with minimal edits showed a spasmodic increase in attendance.

We also noticed that if the article did not start growing during the first month after the update, then there is no need to expect further growth.

Interestingly, bursts are observed after the second actualization of the same article. For example, we had an article "Six ways to attract free traffic to the site." It was written a long time ago, and there were recommendations in it, such as “driving the site away from directories”. It is clear that this is no longer working. Made an article “Attract free traffic in 2018: what methods still work” - described 10 working and 3 non-working methods. At the beginning of 2019, it was updated again - this is the latest edition . URL not changed - left /shest-sposobov-besplatno-privlech-trafik-na-sajt.html. The article shot well, and twice after each update. After the second update, the growth went exponentially:

Thus, if you have articles with good traffic potential and are linked to the year, keep them strictly accounted for and update them when they need them, but at least once a year.

Traffic loss with time

After rapid growth, there is often a decrease in traffic. How strong is another question. For example, in the case of an article about the parsing of search clues, the drawdown was noticeable.

And in the case of the article about the acceleration of site indexation, the drawdown is more like stabilization.

That is, if you have increased traffic after updating the content, and then fell, do not rush to look for reasons - this is a normal process. The main thing is that after the fall of attendance was higher than before the update.

But this is not always the case. For example, what to do if there is such a picture:

When gathering positions, we advise you not to do anything in the first 3 months. Just wait. It happens that the positions are returned. Sometimes triumphantly:

If after 3-4 months the article itself does not come out of the drawdown, update it.

Life hacking: when updating such an article, try using the skyscraper method - analyze competitors from the top and make a new article better in all respects (text length, content quality, optimization).

No reaction

Several articles did not respond to optimization and updating at all. For example, this was the case with the article on vital inquiries .

As we later found out, at the start of work, these articles were in the TOP for the main requests. The revision did not give impetus to the growth of positions in organic matter, since, as a matter of fact, there is nowhere else to grow within the subject matter.

The lack of response does not mean that such articles do not need to be updated. If this is not done, then over time their place will be taken by more recent articles of competitors, and it will be more difficult to return positions. Here, the determining is the assessment of labor costs for the control of such articles. There are resources - pull the maximum out of the article.

Updated articles - what's next?

All around. Update is a continuous process. It does not work and do not rest on its laurels. Content changes every day. That Adwords in Ads renamed, then Direct new interface rolled out, then Google+ was closed ... And every time you need to correct articles, change screens, or even rewrite everything in a new way. Yes, it's a chore and not as interesting as we would like, but the result is worth it.

This work can be approached in two ways:

1. If you have a solid blog and a large editorial team, you can select a special person who will spend some of his time on traffic analysis and article updates.
2. If you - like most companies - have limited resources for a blog, then you should just put yourself in the plans every six months / year to analyze the TOP articles on attendance and see what needs to be corrected, updated and rewritten.

And may the power of SEO be with you.