Linkplanner case study


Published on

This is a run through of our link planning tool. its a bit like Link research tools , except it's in excel, so its very easy for us to manipulate the data to give us more insight into a set of link profiles.

Published in: Business, Technology, Design
  • Be the first to comment

  • Be the first to like this

No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide

Linkplanner case study

  1. 1. LinkPlanner ProcessHow to analyse and plan your link building
  2. 2. Pick your keywordsPut 50 relevant keywordsinto the share of searchtool.With Google keywordfinder, get local searchvolumes and cost per clickdata.Keywords to getshare of search
  3. 3. The tool scrapes google SERPSThe tool queriesgoogle so thereisnt a ‘filterbubble’ i.e. nopersonalisedcontent
  4. 4. Results are shown on a pivot tableThe formula is:CPC x VOL x Rank X No. ofKeywordsThis gives us meaningfuldata showing who hasgreatest share of searchover a set of keywords.The spammers are doingwell…
  5. 5. Take the top domains you care aboutIt’s important to put the domains in order ofshare of search to make meaningfulcomparisons later on.If you are looking at SERPS, the scraper willtake the top 10 results for a keyword andshow them in order of ranking.Link planner can look at specific URL’s orwhole domains.For this example, were looking at URL’sbecause looking at i.e.,rather than makes the dataclearer.
  6. 6. Why we trust Majestic TrustFlowWe took 400 randomly chosen .uk sites from Majestic’s 1,000,000 top sites by backlink count and got the totalsearch traffic figures from SEMrush for each of these domains. We then got the Majestic TrustFlow and CitationFLowmetrics for the domains and plotted the results. (Organic traffic on a logarithmic axis. The black line is theexponential trend line).Conclusion: This data shows that TrustFLow is a predictor of traffic volumes 75% of the time, whereas CitationFlow isa predictor of traffic only 61% of the time. This is why we use Trustflow as our core site metric.
  7. 7. Lets start our analysisWere looking for patterns that give us guidance on what we should do if we want to rank in the ‘casino’keyword landscape.Our underlying assumption with spammers who rank is that they know how to game Google’s algo.Unfortunately, we have lots of spammers, so it’s fairly easy to analyse what they are doing differently torank in such a competitive segment.For operators, it’s a case of understanding why those with the biggest share of search, rank, whilstaccounting for what lessons we can take from spammers.
  8. 8. Overview of domainsThis is an overview of domains, biggest share of search at the top. SkyVegas has a lot of site wide links.Conclusions so far: Sitewide links may not be so toxic after all.
  9. 9. Backlink growth over timeGrowth of new links over time. Interestingly the spam site had gone from no links tomost new links in March. The most aggressive link acquisitions come from the must successful spammer and32red.Conclusions so far: Brands (apart from 32red) are far more conservative about link growth than thespammers.
  10. 10. Cumulative growth of linksWhen seen in cumulative view, 32Red have been very busy. They are a huge outlier in this keyword landscape.Amongst the brand sites, is amongst the lower end of the packConclusions so far: 32Red is being far too aggressive on domain acquisition. are in a nice‘safe’ zone.If you agree with ‘safety is in being the pack’ then it sensible to benchmark off those sites that have greatest shareof search and (within reason) least new domains (Paddypower & Gala Casino)
  11. 11. 32RedOops! That lookslike a penguin2.0 hit!
  12. 12. Link growth over time: Non Cumulative, Tabular formOnlinecasinotop did a ‘blast out’ of links and took share of search over a 6 week period.Conclusions so far: Links ‘kick in’ within 4 – 6 weeks. 18 – 25 new domains a month is good for this segment.
  13. 13. Link growth over time: Cumulative, Tabular formThis is historic link data from majestic seo. 32Red have a huge legacy which in theory should help them, but as yousee it’s out of step with what the Google algo wants today. This is why is so interesting.Conclusions so far: Having a big link history isnt necessarily a good thing.
  14. 14. Linking Domains by Trust Flow bandHere we see the referring domains from Majesticseo’s fresh index set out in bands using Majestic’s trust flow. Sites like32red, Skyvegas and Casinobonusnodeposit5 (spammer) have far more linking domains in the upper ends of the scalethan Onlinecasinotop or Paddypower.Conclusions so far: Winning links, from very high trust flow sites is great, but isnt a precursor to ranking.
  15. 15. Absolute linking domains x trust band & share of searchThis is just another way of presenting the data. Casinobonusnodeposit5 has been very busy and seems to have taken theopposite strategy to the other spammer onlinecasinotop.Conclusions so far: Volume of links isnt that important.
  16. 16. Referring domains by percentage / bandThis view of the data shows the percentage of linking domains per trustflow band. There is a common perceptionthat higher quality linking domains is better, but this does not seem to bear out here.Conclusions so far: There are no clear patterns here…
  17. 17. Linking domains by percentage & Trustflow BandWhen seen inthis way, thebrand sites seemto have verysimilar profiles.
  18. 18. Linking domains percent difference from averageHere we take the average number of linking domains and look at how much a domain deviates from the averageper trust band. It makes it easier to see which domains are strongest per band.We have taken out onlinecasinodeposit5 out because its link profile is very distorting.There is a definite pattern showing the biggest share of search phrases have less TrustFlow per band than the others.Conclusions so far: Spending a lot of money to ‘get ahead of the average’ isnt a clever tactic for 32red. The 2 winningsites prove this point.
  19. 19. Conclusions so far on ‘links’The common thread with spammers is that they can rank from nowhere in about 6 weeks. They vary in the number oflinks used and their quality i.e. onlinecasinotop uses relatively few and they are fairly low trustflow,onlinebonuscasinodeposit5 uses far more links and of a far higher trust flow than almost anyone else, but does not winon search share.The common thread with the operators is that volumes of links and high trust flow isnt that important either. This ispositive because it shows that with good planning you can rank on casino phrases with relatively few links per month i.e.18 – 25 good links.Next is anchor text analysis.
  20. 20. Anchor text analysisWe assign aclassification to theanchor text per domain.This allows us to makesense of anchor textand inbound links.The process typicallytakes about 15 minutes.‘Code’ is theclassification type weuse for each phrase.
  21. 21. Anchor text breakdown by domainThere is a hypothesis that too much exact anchor text is a bad thing, yet have more exactthan any other operator, but have balanced it with the greatest amount of brand anchotext of any operator.Conclusions so far: Anchor text with money words are ok, as long as you balance it out with brand words
  22. 22. Anchor text split outAll the ‘winning’operators have similaranchor text ratiosexcept for 32red whoappear to have a lot of‘other’ anchor text.Conclusions so far:having lots of randomanchor text to looknatural does not helpyou.
  23. 23. Consolidated anchor text split outOperators typically have 55% of their anchor text as ‘brand’ or brand variant i.e. their URL as a link. The variable isthe ratio between ‘keyword’ and ‘other’.Conclusions so far: Having a lot of ‘noisy’ links with random anchor text isnt going to help you much. It’s best tofocus on a 60% combination of brand and 40% combination of relevant keywords.
  24. 24. Consolidated anchor text breakoutSame data as before.Interestingly, if you discount‘other’ keywords, theweighting on brand is huge.Conclusions so far: If you haveto sit on one side of the fence,its on ‘brand’The anomaly is where youhave exact match domainsand so all your anchor textbecomes exact match.
  25. 25. Relevance of anchor text and title tagsBy looking at the number oftimes a keyword is found inthe title tags of pages for allinbound linking pages, wecan assess the relevance oflinking pages.Sites with a high instance ofkeywords in the title tagsshow high subject relevance.Conclusion so far:Relevance of content fromwhere a link is placed is notimportant from analgorithmic point of provesthis
  26. 26. Overview conclusions from LinkPlannerOverview comment. Either the Google algo is very comprehensive or very basic. Spammers prove that very diverse strategies getthem ranked and whilst you may think ‘the algo will get them’, these sites will generally rank until they have a manual penalty. Thisproves they understand how to game the Google algo.When brands are involved, its about a combination of sustainability and prominence i.e. ranking and not being penalised by Google.Since Google are vague about what exactly gets you a manual or algorithmic de-ranking, link strategies need to be conservative. Italso suggests that building your own private content network makes a lot of sense in certain contexts when both the risks andrewards for brands are so high.Key findings:• There is a 6 week lag between dropped links and rankings.• You can aggressively link and rank, but run a high risk of manual penalty• Anchor text mix is fairly important with a weighting towards ‘brand’, which may explain why exact match domains still do wellbecause they get relevant anchor text and a ‘free pass’ on excessive use of exact match anchor text• Content relevance of linking domains does not seem that important• Which country you get links from does not dictate where you rank. Just set your territory in webmaster tools.• On page relevance is very important.• Dependent on strength a single page can rank for up to about 25 similar topic, core keywords.
  27. 27. Plan of actionFor brands, sustainability and relevant share of search are key. So a sensible rule to employ is ‘plausible deniability’. Thismeans when your site is assessed, you will pass a ‘spam’ test.Suggested actions:Don’t have ‘cheap links’, they seem to be the target for Penguin 2.0. Spammers hack into sites for links, you have to getcreative, stay within your brand ethics and have plausible deniability.Look at how to win links from strong non casino sites i.e. chambers of commerce, charities and so on. Do online PR /sponsorship deals to drum up attention and thus links. Topical relevance is not that important.Work within the natural limits of your segment. For ‘Online Casino’, it 18 – 25 new linking domains a month with a MajesticTrustFlow of above 10 on your links.Keep a heavy weighting on brand anchor text (60%+) and mix up exact match across at least 6 phrases. It’s important to have‘natural’ looking anchor text that is in line with other competitors in your segment.
  28. 28. Appendix• Screengrabs of ranking history for domains in question
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35. casino.paddypower.comOur Searchmetrics account doesn’t give subdomain data, so we are using SEMrush datahere.
  36. 36.
  37. 37.