There were many, many processes that worked without any basis – we need an SEO strategy. Already within the framework of the optimization plan, semantics were quickly collected and critical problems were worked out. Then they decided to move the catalog — they left the flat structure. We formed a backlog of tasks, where analytics and fundamental reports are the key ones. We decided on the tools — there will be an interesting case on the example of DATAFORCE from SEOWORK.
Yandex Metric and Google Analytics desktop tools Ionica, but still they are not designed to load huge amount of data. Services, firstly, begin to slow down, and secondly, there are no filters for the collected segments. Due to the inconveniences of analytics, these tools turned to Google Data Studio and SEOWORK.
In GDS, dashboards were built on which pages were filtered according to the necessary queries, and also 3 fundamental reports were created: traffic, site search and Google Search Console (the latter service is often used by project designers). GDS dashboards are quite simple, but their advantages include processing a large array of data without lags, filtering entry points, searching for pages outside the structure, as well as fixing sudden changes (increase/decrease). In Printbar, the data is often considered: we compare 7 days to the previous 7 days and look at the dynamics.
Google Data Studio — already filtered data
About 10 thousand clicks in Google accounted for “controversial” queries. How was the work with keys 18+ optimized?
Removed all “adult” layouts from the children’s category
Removed recommendations of products with prints of 18+ categories for children
Children’s products are not shown in “adult” products
Result of semantics optimization 18+
Thus, the impressions were reduced to about 1000.
SEO Analytics Tools
As mentioned above, 3 types of reports are compiled in Google Data Studio. Along with this tool, Printbar uses Power BI for organic traffic. PBI is more convenient than GDS, but the first one needs good support and infrastructure, because data from BigQuery is tied here, and this already requires certain skills from the analyst. Additionally, the project collects data on positions from Serpstat, basic text analytics in Searchlab and occurrences in texts in Labrika.
We get most of the data from SEOWORK.
The selected software allows you to collect reports that are important for the work of SEO, marketing and design departments. They illustrate the interests of users that are converted into specific tasks. In other words, the designer sees a request for which it is worth drawing a new layout.
In addition, the dynamics of the project are always in sight. Information about the growth or sharp drop in categories makes it possible to respond to circumstances in a timely manner and without traffic losses.
It has become easy for the SEO Department to work with new/old collections in SEO-admin:
simple import and export
API for feeds
managing linking blocks
template engines for all meta and texts (for listings and cards)
easily build segments for SEOWORK
For a site with more than 500 thousand pages, and even more so in 1+ million, such detailed analytics is extremely important, because a lot of garbage accumulates at the entry points
It would seem that what prevents Printbar from generating a million queries, creating the same number of pages and doubling or even tripling the profit? Alas, it’s not that simple.
Difficulties with semantics
Where to look?
eCommerce of this kind includes anything, including memes, phrases, zodiac signs, holidays, cities and an infinite number more. However, in parallel with “everything” there are also licenses — you can’t just take and print Spider-Man.
In general, the wordstat of the guys from Printbar is very peculiar. It seems to display everything, but it does not always have time. For example, today there will be a release of a track by a popular artist, in the evening it will become a hit, and tomorrow everyone wants to find trending prints. Because of this, the traffic share of the project is 50% listings and 50% product cards. Hence the important conditions:
the name of the card affects a lot
, the presence of the card in the index is necessary
, there is an irrelevance of the card (mainly in Google)
Report from Google Data Studio + Yandex Wordstat
Wordstat does not always reflect reality/relevance and does not understand the intent. For example, requests about merch with the singer “shaman clothes” are not taken into account by Wordstat. Thanks to the internal site search report, only relevant project queries are taken into account. Noticing the real demand, the team creates a new print and a new product card that the user can buy.
How does Printbar look for semantics?
Internal search, webmasters — a key tool
Competitor parsing and issuance
Keyso, moab, alphabetics, overlay services
Info sites for movies, games
Sites with pictures
Memes, situations and other creative
The found queries are checked in the tools for the mass nature of the desire and, with appropriate indicators, it is launched into operation.
How to collect?
The project has its own feature. In order not to divide a large array of data into groups later, the team immediately forms them. All collections are collected through a scraper, an elementary coupling is created, for example, [by age — 27 years: 27 years of T—shirts] or [Rap — 2Pac: 2Pac T-shirts], etc. – all this is parsed, frequency is checked, and as a result, semantics is ready to be uploaded to SEOWORK and other tools. This method can be used in any subject (metal products, industrial equipment …).
Generzhka: we take “nature”, collect the entire dictionary, normalize it in py7.ru/tools/norm/
We combine it with product types, commercial and thematic masks
Parsim hints, wordstat, mutagen, overlay, word-keeper
Parsim rel. pages (suddenly there is already, but there is no top)
Through the rel.pages, we parse more in moab, keyso
Parsim info sites of any subject: movies, music, anime, games
Parse the necessary fields of the type: author, director, actor, hero