Link building case study: a link network plan for a furniture industry website

This text is a link network proposal we prepared in 2009 for Mebelskol, a company in the furniture industry, and published on the blog in 2012. At the time the company website supported sales and brand image for a local business that fully equipped schools and offices with furniture and teaching aids. In 2026 we are leaving the proposal in the shape it was written and adding an assessment: which activities Google now treats as spam, which are still ordinary marketing, and how the same goal could be reached without the risk. The 2009 document contains no campaign results, so we do not give any here.

#Mebelskol website history before the link network

The company has existed since 1993. It continues the tradition of Mebelskol, the nationwide school supply network that held a monopoly in this market before 1989. Long cooperation with schools gave it an established brand. Easier internet access, schools becoming more aware of the competition and a growing number of competitors nevertheless forced it to cut costs radically and move some departments out to external providers.

The website was launched in 1998. It was meant to help potential customers find the offer and to let existing customers follow new products without calling the company. Until 2006 it was a business card site with information on the main areas of activity and a few dozen products. The code was sloppy (among other things it blocked search engine robots), and the design did not match the company’s visual identity, with unusable colours and weak product presentation. In 2006 a new site with 2000 products was prepared. In 2009 it was rebuilt for search engines and usability, with the new version scheduled to launch on 1 February 2010.

#SEO link network goal for a company furnishing schools and offices

The management board noticed that the website drove sales: in 2008 it accounted for 40% of turnover, even though the company had a group of regular customers strongly tied to it offline. Before that, the website’s share of revenue had been treated as fairly marginal. In 2009 the decision was made to expand the link network, optimise the site for search and increase brand recognition.

We had to agree with the company on what goals the website should serve. A conversion was contacting a branch to order products. On top of that came the SEO strategy: the target group, a list of activities for building the link network and a list of activities on the site itself. This was not simple. The company knew its offline customers well, because they were regulars. Online customers still had to be identified and described.

The assumed target group:

• technical and administrative directors of public, educational and cultural institutions and of offices,
• people responsible for public procurement,
• residents of towns with more than 10,000 inhabitants who buy online rarely or not at all.

The overriding goal was to reach more new customers across Poland in the company’s key areas of activity. Higher positions in search results were supposed to turn into conversions, meaning contact with the head office and a placed order.

#How the main site was prepared for long tail in 2009

Before the link network got going, the site itself was put in order:

• Hosting in Poland. The audience was Polish, so latency and loading speed mattered.
• WordPress as the engine. The choice fell on a free system with a very large number of installations and extensions, assessed for security, room for growth, flexibility and update frequency, with a Polish language version.
• Competitor monitoring. SEM, public relations, keyword choice and on-site activity of competitors were analysed. Based on this, a plan was drawn up that was meant to give the site more visits than competitor sites within 12 months.
• Keyword research for long tail. The main source of search traffic was supposed to be phrases of 2 to 5 words. The list was built with Google External Tool, SEOmoz tools and Keyword Elite. Landing pages were prepared for the phrases with a refined title, meta tags, image descriptions and content formatting. The long tail covered queries about specific products and their features, because such a user knows what they are looking for and is close to ordering.
• Site map with dependent subpages. Based on the phrase list, a structure was created in which subpages were extended with further ones that captured traffic from long queries, for example nauczanie zintegrowane pomoce dydaktyczne (integrated teaching aids for early education) or anatomia człowieka modele narządy wewnętrzne (human anatomy models of internal organs).
• Short phrases. For 1 to 2 word queries, combinations of brand, location and category were chosen: globusy (globes), meble Warszawa (furniture Warsaw), wyposażenie szkół (school equipment), krzesła Nowystyl (Nowystyl chairs).
• Code and indexing. The code was optimised for speed and for easy crawling and indexing. Forms, general instructions and pages unrelated to conversion were excluded from the index, and links to unwanted pages were marked rel="nofollow".
• Sitemaps. A navigational sitemap for people and a sitemap for robots were created and submitted to indexing systems and social services (Friendfeed, Google Webmaster Tools, Twitter, Blip). A robots.txt file was prepared.
• Error pages and navigation. Error pages and navigation elements were refined for usability.
• Internal linking. Only after the internal link structure was settled did building the external link network begin.

On the site itself several navigation paths were also planned (menu, breadcrumbs, sitemap, tags, internal search results, related products, favourites), along with typography and readability fixes and copy written in plain language. On top of that, material users like to pass on: unusual products and promotions.

#What from this part still holds up in 2026

Most of it. A structure built around long tail queries, a tidy index, sitemaps and internal linking are still the foundation. Two elements need correcting.

The first is emphasising keywords by repeating them densely. Google’s policies define keyword stuffing as filling a page with keywords or numbers in an attempt to manipulate rankings, and give as an example blocks of text listing cities and regions the page is trying to rank for (Google Search Central, spam policies). A landing page for a long tail phrase should answer the question, not repeat the phrase.

The second is rel="nofollow" on internal links. Google now says plainly that for links within your own site you should use a disallow rule in robots.txt if you do not want the crawler to fetch a given page (Google Search Central, qualify outbound links). The nofollow, sponsored and ugc attributes are for describing outbound links.

#What link network was planned for a furniture store

External activities in the 2009 proposal:

• Themed shops where new and existing customers can order directly. This was meant to take load off the customer service department.
• Regular mailings to registered users with promotion details and a discount code, for more frequent contact with the brand.
• Subpages closely tied to product categories, forming a link network that links to each other and to the target site.

As part of the link network the plan included:

• regional offer sites, such as meble.warszawa.pl, supporting virtual offices in different cities with local phone numbers forwarding calls to the head office. The assumption: customers trust a company from the same city or region more,
• sites with a narrowed product range, all on different IP addresses and different scripts, varied in content, anchors, link placement and code,
• a site with a niche offer for private schools with equipment imported from Norway and the United Kingdom: more expensive, more durable and better-looking products with a higher margin,
• sites focused on the most effective long tail phrases in the industry,
• website directories with an overview of the company’s offer and references to manufacturers in the industry.

In addition:

• connecting some of the link network sites to link exchange systems (SWL). The main site was not connected, because that was risky: a filter imposed on a network site was not supposed to carry over to the target site,
• listings in paid and free directories, industry business lists and local business directories,
• commissioning a few dozen industry articles through the zlecenia.przez.net portal and publishing them on presell pages and on citizen journalism portals such as sfora.pl and wiadomosci24.pl,
• placing the company’s RSS feeds in a few dozen aggregators (technorati.com, friendfeed.com). The proposal itself noted that links from RSS are often copied by spammers and not very valuable,
• social network profiles connected to ping.fm,
• video channels with manufacturer films, the company’s own assembly instructions and industry news,
• a free ebook on the ergonomics of office furniture with advice on healthy use,
• PPC campaigns in the periods of highest turnover,
• tracking competitors’ backlinks through Yahoo! Site Explorer and getting links from the same places,
• publishing manufacturer news on the company site,
• cooperation with industry sites such as meble.pl: ads and sponsored content on those sites and in their newsletters,
• links from the sites of manufacturers, distributors and companies working with Mebelskol,
• image-building sales on auction sites.

Offline activities:

• free training for sub-distributors with materials available only on the company website,
• sponsorship of school competitions with the website address on banners,
• 10 types of product catalogues with a reference to the website on every page,
• staff taking part in industry symposia and meetings,
• stickers with the website address on every item leaving the warehouse.

When published in 2012, the proposal cited two sources: the paper by Baoning Wu and Brian D. Davison on detecting link farm pages (Lehigh University, 2005) and Rand Fishkin’s Whiteboard Friday episode on how the link graph works (SEOmoz). So even then it was known that search engines look for networks of sites linking to one another.

#How Google defines link spam and link networks today

Google’s current policies define link spam like this:

“Link spam is the practice of creating links to or from a site primarily for the purpose of manipulating search rankings.” (Google Search Central, Google Search spam policies)

Among its examples Google lists, among others: buying and selling links for ranking purposes, excessive link exchanges and partner pages created exclusively for the sake of cross-linking, automated programs and services that create links, links from low-quality directories and bookmark sites, links widely distributed in the footers and templates of many sites, and articles, guest posts and press releases with keyword-rich anchor links. Neither the term “link network” nor “PBN” appears in the policies. What counts is the purpose: if the sites exist mainly to pass links, they fall within the definition.

The same page separates advertising links from spam:

“Google does understand that buying and selling links is a normal part of the economy of the web for advertising and sponsorship purposes. It’s not a violation of our policies to have such links as long as they are qualified with a rel=“nofollow” or rel=“sponsored” attribute value to the <a> tag.” (Google Search Central, spam policies)

Google also describes how it enforces these policies: it detects violations automatically and, where needed, through manual review, and sites that violate the policies may rank lower in results or not appear at all.

#Which 2009 link building tactics break Google policies today

The assessment below refers to the policies Google published as of 2026, not to the realities of 2009.

The core of the link network itself carries the most weight. The proposal assumed that a filter on a network site would not carry over to the target site. Google states that its algorithms and manual actions aim to nullify such unnatural links at scale, so the target site loses what the network was supposed to give it, even without a penalty imposed directly on it.

#How Google policies changed from 2005 to 2024

The proposal was written at a time when some of these practices were common. The subsequent changes on Google’s side went like this:

• 2005. Google introduces the nofollow attribute as a way to fight comment spam. It quickly also becomes the recommended way to mark advertising links (Google Search Central Blog, 10 September 2019).
• 24 April 2012. Google announces an algorithm change targeting webspam that lowers the rankings of sites violating its quality guidelines. Among the techniques it lists keyword stuffing and link schemes (Google Search Central Blog, 24 April 2012). We published the Mebelskol proposal on the blog in February of the same year.
• 10 September 2019. The rel="sponsored" and rel="ugc" attributes are added, and all three values become hints rather than a hard exclusion. Google explains that analysing all links helps it better recognise unnatural linking patterns (Google Search Central Blog).
• 14 December 2022. A link spam update using SpamBrain:

“SpamBrain is our AI-based spam-prevention system. Besides using it to detect spam directly, it can now detect both sites buying links, and sites used for the purpose of passing outgoing links.” (Google Search Central Blog, December 2022 link spam update)

In the same post Google writes that as unnatural links are neutralised, rankings may change, because the credit those links passed is lost. Sites with a narrowed product range, designed to pass outgoing links, are exactly that type of site.

• 5 March 2024. Three new policies: expired domain abuse, scaled content abuse and site reputation abuse (Google Search Central Blog, March 2024).

#Expired domains and commissioned articles in the 2024 spam policies

The 2009 proposal did not involve buying domains from previous owners. The policy is still worth describing, because today’s link networks often start with exactly such domains. Google defines expired domain abuse as buying an expired domain and using it mainly to manipulate rankings by hosting content that provides little value to users. In the March 2024 announcement it adds:

“Expired domain abuse isn’t something people accidentally do. It’s a practice employed by people who hope to rank well in Search with low-value content by using the past reputation of a domain name.” (Google Search Central Blog, March 2024)

The same post notes that an old domain used for a new, original site created primarily for people is not a problem. The line is the purpose, not the age of the domain.

The second policy concerns scaled content. Google describes it as generating many pages mainly to manipulate rankings rather than to help users, and stresses that it does not matter whether the content was created automatically, by hand or by a combination of both. A few dozen articles commissioned on a freelance marketplace and spread across presell pages and citizen portals fit this description if their only job was to carry a link.

The third policy, on site reputation, covers third-party pages published with little oversight from the host in order to exploit that site’s ranking signals. Google notes that not all third-party content is a violation: native advertising intended for the site’s regular readers does not have to be blocked. For cooperation with a site such as meble.pl, this means a sponsored article is fine as long as it is written for that site’s readers and the link is marked as sponsored.

#How we would build a link building plan for Mebelskol in 2026

The company’s goal was sensible and would still stand: more enquiries from schools, offices and public procurement departments across Poland, without relying only on regular customers. The route would change.

Stays as it is. A site structure built around queries for specific teaching aids and furniture, technical housekeeping, sitemaps, internal linking, mailings to registered customers, the ergonomics ebook, assembly instruction videos, printed catalogues, training for sub-distributors, sponsorship of school competitions and presence at industry meetings. Google’s SEO starter guide lists social media promotion, community engagement and online and offline advertising as ordinary ways to promote a site, and describes word of mouth as one of the most effective and long-lasting, though slow (Google Search Central, SEO starter guide).

Changes. Instead of sites with a narrowed product range on separate domains and IPs: the same content as sections of the main site, including a separate section for private schools with equipment from Norway and the United Kingdom. Instead of regional domains with virtual offices: location pages only where the company has a real branch or service, with a real address and delivery area. Sponsored content on industry sites stays, but with rel="sponsored". Links from manufacturers and distributors stay, because they come from real cooperation.

Goes. Link exchange systems, presell pages, articles commissioned only to carry a link, directories built for links and a network of sites meant to hide its connection through different IPs and scripts. Each of these now falls directly within the examples of link spam or scaled content.

Gets added. Content that earns links on its own: for example a guide to school furniture standards or a guide for people preparing a public procurement tender for equipment, written for the target group already described in 2009. We cover earning media mentions separately in the article on getting links through media and public relations, and the concept of a link network itself, its types and strategies in the guide on PBN vs link building.

This proposal is a record of what SEO looked like in 2009. More interesting than the tactics themselves is that the goal and the target group have not aged at all, and half of the planned offline and on-site activities are still exactly what Google describes as promotion that follows its policies.