An Ode to My Favourite Advanced Search Operator

For a short, Sunday morning blog post to keep things ticking over, I thought I’d write a brief tribute to my favourite (and often overlooked) search operation, related sites.

I use this search operation all the time. All Google does is throw up sites it thinks are similar to whatever you’re using as your base.

It’s perfect for:

  • Finding overlooked competitors
  • Looking for similar, relevant sites when link prospecting
  • Identifying quick-wins from places similar to where you’ve already had success
  • Throwing up new ideas when your efforts reach an impasse

What’s amazing is how often good quality prospects come up in related searches that haven’t shown up in your previous research (well, for me anyway).

Whenever pitching content at a website, I’ll always make sure I run a related search on the website like if you want to search Web Design Ireland, you need to have search operators to search that site.

The first few sites that show up are usually also good targets for the content, so I can increase the chance of it being taken by pitching it at them too.

I sometimes think the best sites to target are the ones that don’t rank well. As Paddy Moogan says here, they’ve not been already been hounded by other SEOs and tend to be more receptive to your approaches:

This is why Twitter and particularly Followwonk are so useful. (See John Doherty’s post on link prospecting with Twitter for more on this.)

Anyway, that’s my relatively inane tribute to related searches. I’m also looking forward to Think Visibility next week and hoping to get a few more substantial posts on here over March.

Experimenting with Custom Reports in Citation Labs’ Link Prospector

As much as I enjoy link prospect­ing, in my heart of hearts, I know it’s not where my time is best spent.

I work in an agency and my time is bet­ter served on the out­reach and acqui­si­tion phase of link build­ing than it is drum­ming up lists of prospects.

I know the less time I have to spend link prospect­ing, the more time I can spend on the work that cre­ates value. The 20% that gives 80% of the results.

One thing I love about Cita­tion Labs link prospec­tor is that it dras­ti­cally reduces the time I need to spend link gath­er­ing link prospects.

Crudely bro­ken down, here is how I used to break up my time spent link build­ing for indi­vid­ual campaigns.

This is how it looked when I was doing a lot of man­ual link prospecting:

Com­pared to now, with a large part of the link prospect­ing automated…

Less time prospect­ing means more time spent mak­ing con­tact and fol­low­ing up, which increases the chance of build­ing more links.

Tools like Cita­tion Labs link prospec­torBuz­zStream and Ontolo are only ever as good as the peo­ple using them.

It is impor­tant to strive to find ways of mak­ing them work for you. A good tool lets you work smarter, as well as quicker.

It is good to be cre­ative in their appli­ca­tion and chal­lenge them to com­plete tasks that would be impos­si­ble, or highly time con­sum­ing, to com­plete manually.

With this then, I’ve been exper­i­ment­ing and test­ing new ways of using Cita­tion Labs using cus­tom, advanced search operators.

Here’s what happened:

Related searches

I blogged recently about my love for ‘related:’ search oper­a­tor.

Usu­ally, I just man­u­ally run these queries one site at a time, scrape the results and then qual­ify them.

It is par­tic­u­larly use­ful at the begin­ning at cam­paigns when you’re research­ing com­peti­tors. I like to make sure I’m aware of every con­ceiv­able com­peti­tor to the site I’m work­ing on.

Usu­ally, the more obscure, the bet­ter, as they’ll have back­link pro­files that haven’t been mined by other SEOs.

I was curi­ous to see what would hap­pen then if I run these queries in bulk.

First then, I set up a cus­tom report…

For the sake of this exper­i­ment, the niche I was research­ing was box­ing and the key­word I chose was box­ing gloves.

I’ll start by tak­ing 10 com­peti­tors and putting them next to the related search query for my search:

My hunch was that the related sites here wouldn’t tell me much I didn’t already know, as I was see­ing sim­i­lar results in all the dif­fer­ent vari­a­tions of ‘box­ing gloves’ I searched for.

I took a sam­ple of urls from dif­fer­ent search queries here, as opposed to just the 10 that appear for the term box­ing gloves, to try make sure I got a bit of diver­sity, whilst still being relevant.

I also think it is impor­tant to find out who your client iden­ti­fies as their com­peti­tors, as they may not actively be opti­mis­ing their site for the search engines.

If they do come up with some­one I’ve not already found, I’ll always throw that site in.

My trep­i­da­tion here was def­i­nitely a bit mis­placed, the search yielded 225 domains. As you would expect, there was a bit of junk in here (MySpace, Face­book etc.), on the whole though, the sites it brought back were rel­e­vant and useful.

Although they weren’t the crème de la crème of the com­pe­ti­tion but this was a good thing. They were decay­ing web­sites but, the point is, they have rel­e­vant back­links I oth­er­wise wouldn’t have found.

If I was being thor­ough here, I’d run another cus­tom related report on the new sites the pre­vi­ous report had identified.

I’d then com­pare the two lists using Ontolo’s Fil­ter Prospects tool to see if there were any more new related domains.

I would then have a com­pre­hen­sive list of sites com­pet­ing with my own. I’d down­load every back­link these sites have and then review to see which might be worth try­ing to accrue.

If you want to pull out all the stops in com­peti­tor link analy­sis, then run­ning a cus­tom related search in link prospec­tor should def­i­nitely be part of your process.

Using tildes

Next up I exper­i­mented by using tildes. Tildes (~) is the syn­onym operator.


(If you want to read more, Gar­rett French’s Guide to the Tilde is a good place to start.)

Again, I used box­ing in my exper­i­ment, just because I like boxing.

First of all I defined the oppor­tu­nity I was look­ing for. In this case, it was guest posting.

To keep this exper­i­ment sim­ple, I’m just used the head term ‘fitness’.

One thing Gar­rett under­lines in his post on tildes is that the impact on using them is diver­sity.

If you con­struct your queries to include all vari­a­tions (which I would advise at least test­ing) and aggre­gate your SERP results, you clearly get a far greater diver­sity of domains. This obvi­ously doesn’t mean that the results are more qual­i­fied, just that there’s more diver­sity with less think­ing on your part about what prospect­ing phrases to use.

What I want then, is for link prospec­tor to run the tilde oper­a­tion quickly and bring me a diverse set of domains, with

My three oper­a­tors were:

~fit­ness –fitness

I popped these into Ontolo’s Query Gen­er­a­tor, which shows 132 foot­print queries that could yield good results.

Nor­mally my next step would be to go through these indi­vid­u­ally and use the scraper tool. Need­less to say, this is tedious time con­sum­ing. With the link prospec­tor tool, I just down­load these queries as a csv and copy and pasted them into a cus­tom report.

When it came back I’ve got the results from the 132 tilde queries above in about, 4 min­utes, mean­ing I could move quickly onto the qual­i­fi­ca­tion stage and more impor­tantly, the out­reach :D

My process from here on in is this:

  • Export the paths to csv (I like to see the path, rather than the domain, so I can work out why it has been identified)
  • Next I want to trim the fat. I order the results by PageR­ank and cut out any­thing above 5. These are nor­mally too high pro­file or sites like Face­book and Twitter
  • Then it is just a case of going through the nor­mal qual­i­fi­ca­tion process using Ontolo’s prospect reviewer


Ulti­mately, the Link Prospec­tor is a time sav­ing device that runs mul­ti­ple search queries for you . It allows you to weight your time onto the business-end of get­ting links.

The point of these exper­i­ments was not so much to show the qual­ity of results, but the ease of which run com­plex search query com­bi­na­tions. It feels like a math­e­mati­cian being given a cal­cu­la­tor to use when you’ve been pre­vi­ously laboured with an abacus.

I’ve only really tipped the ice berg here. Would like to see other peo­ple exper­i­ment­ing with cus­tom reports, see their process and see what kind of results they get.