Penguin 3.0 and Links Bought Years Ago

An old invoice for directory submissions

An owner going through old files finds invoices from a marketing company that once charged a monthly fee to submit the website to directories and article sites.

Last Friday Google began rolling out a new version of Penguin, the part of its ranking systems that looks for manipulative links. Google says the rollout is gradual.

Penguin was built to find links bought this way. They did not expire with the contract; they sit on those pages until someone removes them.

What Penguin looks for

Google treats a link as a sign that someone vouched for a page. Links that were paid for, swapped or placed by the owner vouch for nothing; Google's guidelines call them unnatural.

Penguin looks for the patterns such links leave: directories nobody visits, articles written only to carry a link, and anchor text repeating the same keywords, usually a service and a city.

It runs as a periodic refresh, not constantly. A site it affects stays affected until the next one, and the last came about a year ago, so cleanup done since is only now being weighed.

What old paid links look like

In a download of the links to your site, the ones bought years ago tend to resemble each other:

• Directories that list anything for a fee, on pages of unrelated businesses.
• Article sites carrying short pieces written to hold a link in the author box.
• Anchor text made of your keywords rather than your business name.
• Footers and sidebars of unrelated sites, repeating one link on every page.
• Networks of thin blogs linking to each other and to paying customers.

Genuine links look different: a supplier's dealer page, an association's member list, a local paper's story. They tend to use your name, and exist because someone chose to link.

A message, or silence

A manual action is a penalty set by a member of Google's staff, and it arrives as a message in Webmaster Tools under Manual Actions. Fixing one ends with a reconsideration request.

Penguin sends no message. It is automatic, there is no request to file, and the only sign is a change in search visits that lines up with the date of a refresh.

Check Manual Actions first. If it reports no manual webspam actions, any link problem the site has is a matter for Penguin and its refreshes.

Reviewing the links to your site

In Webmaster Tools, the Links to Your Site report lists the domains that link to you. Download the latest links and sort them by domain, so each site appears once.

Go through the list with the old invoices beside you. Mark each domain as genuine, paid or doubtful, and open a page from every doubtful one before deciding.

The download is a sample, not a complete record, but it is the right place to start: these are links Google has already found.

A worked example

A retractable screen installer with two vans fits disappearing screen doors and motorised patio screens in homes around Greater Victoria. Six years ago it paid a marketing company monthly for directory and article submissions.

In the spring a new office manager downloaded the links and sorted them by domain. Most came from directories and article sites, nearly all with the same anchor text: retractable screens and a city.

Mixed in were links worth keeping: two manufacturers' dealer pages, a home show's exhibitor list and a builder's page naming its trades.

She wrote to every site with a contact address and logged each request. A few links came down; most sites never replied. In June she uploaded a disavow file listing the remaining directory and article domains.

Last Friday's refresh is the first since that work. The approach follows how Penguin reads links: what still counts toward the site is what others chose to give, and the bought links are set aside.

Removal first

Write to the directories and article sites asking for the listings to be taken down. Many are abandoned and never answer; some remove a link the same day; a few ask for a fee.

Keep a record of each request and its date. It shows the effort if anyone ever asks, and tells you which links remain for the next step.

What a disavow file is for

A disavow file is a plain text list of links you want Google to ignore, uploaded through the Disavow Links tool in Webmaster Tools. Each line names a single page or a whole domain.

It is for links you know are manipulative and cannot get removed. Once Google recrawls those pages, it generally treats the links as if they did not point at your site.

Disavow whole domains when the site is a directory or a network. Listing single pages tends to miss the copies of your link elsewhere on the same site.

When disavowing is the wrong tool

It is the wrong tool for links you are merely unsure about. Disavowing a genuine link throws away something that helps, and the file cannot sort good from bad for you.

It is also wrong when a drop in visits has another cause: a redesign that lost pages, a quiet season, a competitor's new site, or thin pages of the kind Panda looks for.

And each upload replaces the previous file, so keep one master list and add to it. Uploading only this month's domains quietly withdraws every earlier entry.

The counter-case

A business that never paid for links, whose links come from customers, suppliers and local press, has nothing to clean up beyond a yearly look at the list.

Every site also collects odd links from spammy pages it never asked for. Google generally ignores those, and they are no reason to fill a disavow file in a panic.

Where links were bought, though, waiting for them to be forgotten does not work. They stay where they were placed until someone has them removed or tells Google to ignore them.

Going through the old links

1. Find the old invoices for link building or submissions.
2. Check Manual Actions in Webmaster Tools.
3. Download your links and sort them by domain.
4. Mark each domain genuine, paid or doubtful.
5. Ask for paid links to be removed, and log it.
6. Disavow the domains that remain.
7. Keep one master file for every later upload.

Step four is where judgement matters most. A genuine link wrongly listed is a loss you never see, so open each doubtful page, or ask someone who reviews links regularly to help.

Keyword-rich links in releases sent to hundreds of news sites are treated the same way, as explained in a press release sent out to get links.

Frequently asked questions

West Coast Media Solutions Inc. provides web design, web development, hosting, digital marketing, and business consulting to organisations across Canada, drawing on more than twenty-five years in the field.