The Link Builder’s Guide to Polling

By Steve Ollington Published April 26, 2013 



Polls can get links. They can get a lot of links, from great places… from niche blogs to national media.

A poll should be looked at as a tool, and if it doesn’t get links, it’s not because of the tool; it’s because of how it’s used. Just like how some infographics get links and some don’t, some articles, some videos, or pretty much some everything! Polls are no different; they’re still content so the same rules on content quality apply. Just like any other content that gets links, it has to be unique, original, and interesting. But… polls also have strict credibility standards to meet if they’re to get published (and get links).

What is a poll?

Before going any further I should quickly point out that a poll is essentially a survey. The difference being that a poll aims to give a ‘snapshot’ of widespread opinion, rather than something in-depth and detailed. For example, a customer survey may include requests for feedback such as “How can we improve X?” whereas a poll will have just a few simple questions with multiple choice answers, and is designed to be representational of a group of people, of any size or type, such as a nationality.

Polls can get links!

Obviously not all polls are going to have the same PR success. However, the example below is something anyone could have run on a smaller scale if they’d have thought of it, and it still would have been likely to achieve some good coverage, based on the contents of the poll.

On 2nd April, the PPP (Public Policy Polling) released data on weird stuff that people believe in the USA.  The poll showed that high numbers of people in America believe things that many other people would consider as quite unbelievable. Essentially it shows a higher proportion of conspiracy theorists and less evidence based critical thinking than might be expected. For example:

  • 15% think the medical industry and the pharmaceutical industry “invent” new diseases to make money.
  • 21% of voters say a UFO crashed in Roswell, NM in 1947 and the US government covered it up.
  • 37% think global warming is a hoax.

That’s a lot of people believing some rather strange things, so this poll doesn’t need to be sensationalised; the results are sensational on their own merit.

On 3rd of April, I noticed it as it was covered in The Guardian, then I saw it getting coverage elsewhere too. This was just part of the view on 4th April in the new OSE feature ‘Just Discovered’:


I’m sure most will agree that some of the sites on that list are desirable places to get links. And that’s without all the links from niche bloggers, etc… of which I’m sure there will be many.

As of 16th April, Majestic SEO is showing 5,000 links from 701 referring domains.


Credibility: Use a polling agency

Lots of polls (and surveys) fail to get publicity and links because they lack credibility and/or interesting content. The example above has interesting content, but had it been conducted through an online survey by some random person with no credentials and specified criteria then it wouldn’t have got all that coverage. This is why you use a proper market research agency (a pollster) for your polls.

If you’re a big enough brand then maybe you can run a poll and get coverage yourself, using the credibility you’ve already built up with public perception and the media. However, most brands need to hire a professional, reputable pollster to run their polls for them, because that means the poll results are more credible, and so more worthy of publishing.

There’s a big difference between a poll on Facebook with 20 respondents, and a poll through a market research agency with 2,000 respondents.

The issue isn’t just one of sample size (although that’s clearly a consideration), it’s also about ensuring that your respondents are representative of the general population.  For example, are your respondents from diverse backgrounds, geographically representational, of differing ages etc? Or are they all fans of yours on Facebook?

The latter is unlikely to provide a truly representative sample not least because they’re all ‘fans’ of your brand, and so the results of the poll would be biased in your favour.

Also, there can be issues with the questions you ask, affecting credibility of the results. If a question is not properly phrased, it can influence the results in one direction or another. Here is an example of a poll which has been roundly criticized due to the interpretation of the results not being in tune with the phrasing of its questions:

Recently the Church of England ran a survey asking:

‘Irrespective of whether you currently pray or not, if you were to pray for something at the moment, what would it be for?’

Based on the results of this question the C of E then claimed that 4 out of 5 believe in the power of prayer (as 81% of respondents gave some kind of an answer). As you might expect there were many issues highlighted with this question, and some referred to the results as nothing more than spin.

One problem with the question is that it uses “irrespective of” – i.e. regardless of if you believe or not. Richard Dawkins pointed out that you could replace prayer with a magic wand and get the same results; it doesn’t mean 4 out of 5 people believe in the power of magic wands.

Imagine you were asked ‘Irrespective of whether you make wishes or not, if you were to rub a lamp and a genie granted you a wish for something at the moment, what would it be for?’ If you answered “World peace”, or “To win the lottery” to that question, does that mean you believe in the power of genies wishes?

Even the UK Census has a similar problem. So much so, there’re whole campaigns being run to change the phrasing of some questions.

Good polling agencies will ensure the questions aren’t leading or biased, will only run with polls that have publicity potential (so can help mould polls correctly to do so), and they might publish it themselves, which often the right people monitor.

What I’m saying here is that if you use a reputable polling agency like YouGovIpsos MORI, or ComRes (for UK anyway) then they’ll ensure the right number of respondents, a diverse audience, and properly phrased questions. That they’re known for doing this means that if they produce your poll, it’ll be taken as more credible, therefore good enough for publication where it otherwise might not have been. So using a trusted market research agency can make the difference to whether the poll is taken seriously enough to publish or not.

Making the poll interesting

The topic and questions have to be interesting. This can be done no matter how uninteresting you might think a product or service is… For example – Sky Scanner ran a poll on which seat people prefer when flying.

A quick search for some of the text in the poll, “Anecdotally some passengers seem to opt for the middle section near the wings”, shows plenty of coverage. There’s some links in there too.

Of course, this leads on to the option of making an infographic, which may encourage more publishing, as it’s a good way for webmasters to provide data visualisation to their readers.


However, in this example there’s more link opportunities than there are existing links. This could mean more links for Sky Scanner if they ask the publishers nicely. More on this later.

The main thing that needs to be thought about here are the core points of the results, the headlines that might be used for those results when published, and the reactions those headlines could provoke. Do the results support or challenge pre-conceived ideas? Do they trigger emotion? Be it shock, surprise, annoyance, sadness… as long as it’s something whether positive or negative, then there’s some potential.

The polling agency will likely need to rewrite your questions to ensure they’re not leading, but have a think about what kinds of questions to ask, based on the results you think would interest people.

The good polling agencies won’t just run anything. They won’t do a poll on something too inconsequential. They might help you build something if you contact them and ask for advice, but have some ideas ready to run through with them. Think about the possible outcomes of the poll, and the responses those outcomes might elicit. Would you share the responses on a social media platform? Would your friends and colleagues? Can you imagine anyone wanting to link to the poll results page because they think it might be of value to their website visitors? These are the questions you must ask yourself when deciding on what to do a poll about.

But how do you go about deciding what to create a poll on in the first place? Well there’s a few different ways:

1) Look at the news. Depending on your industry you might find stuff getting national media coverage which is relevant to you, but most likely it will be a case of looking at industry related news. Is there anything trending within your industry that you could run a poll on? Have a look. Do some searches in Google News with industry keywords.

2) Check Social Media. Are there discussions on social media networks, blog comments, and within forums about anything related to what your company does? Read some. See what people are saying. Do any of the discussions contain a good poll idea?

3) Complaints sites/review sites. What are the problems within your industry? Are their common issues that could be reflected with the results of a poll? Read review sites, find complaints about anything relevant to what you do.

4) Just think. Seriously, just sit there and think, like you would do with trying to come up with ideas for any other kind of content. Look at stuff around your office or in folders on your laptop. This might sound ridiculous but honestly just playing with random ideas can work. It probably doesn’t always work, and you’ll probably have a bunch of daft ideas before you think of a good one but still, you might think of a great one.

5) Don’t force it. At the risk of totally contradicting everything I just said above, let the ideas come to you. The thing with ‘finding’ something to poll about is that it’s restrictive, and that may show in the final outcome. The best idea I believe I’ve had was when I wasn’t looking for it. I just spotted a headline of an article; I disagreed with the headline and so clicked to read the rest of it. By the end of it I’d found flaws in the argument to my mind but I had to prove those flaws somehow… a poll would prove those flaws (I’m still yet to do this poll, so I haven’t proven squat yet, and that the poll doesn’t yet exist is why I said it’s the best idea I ‘believe’ I’ve had. I can’t know for sure until I do it and see the results).

How to ensure the link building success of the poll

Some companies have done the above. They’ve got their poll results mentioned on news sites and lots of other places. However, often news sites will simply state “A recent poll shows…” and/or mention the company by name, but won’t actually link.

As such you’ll need to be prepared to some extra work to secure links.

You’ll first need to find where your poll results have been published. You could search on Google for unique phrases with quotation marks “Like this phrase containing this link for further information on search operators”, you could use some social monitoring software, or you could use copyright software like Copyscape. Either way you’ll be looking for content that matches your poll results. Once you’ve found instances of your poll results, you then need to contact the relevant person to politely request a link.

These sort of requests ought to be handled carefully – after all you don’t want to irritate the person who’s given you coverage! I’d recommend sending a short email thanking the writer for covering the story, along with a polite suggestion that their readers might like to see the original source of the data, or any extra data that may be provided on the page in question. Include a link to the source that they can easily insert. Don’t push them too hard, they’re busy people with other work to do. Also, don’t expect 100% success – some won’t link to you. Don’t get silly at this point – let it go. They could be a valuable contact in the future and it’s not worth spoiling things for the sake of one link.

How much do polls cost?

Prices range from about £250 – £1,000 per question, depending on the pollsters used and the number of questions asked. So it’s not a ‘cheap’ option.

In reality you may not get the kind of coverage and link equity the ‘strange beliefs’ poll above did, but there’s still the potential for great value there. A good poll gives you content that’s link-worthy, and a great PR resource.

How do you find a credible pollster?

Okay so there are a lot of market research agencies out there, with varying levels of credibility. You’re going to need to do some research yourself into which to use. Try this:

1) Look for some existing polls that have been published (Just Google some different keyword phrases with ‘poll’ or ‘survey’, and try within the Google News vertical).

2) What site did you find the poll on? Is it a good site, maybe even a national media site?

3) Has that poll got coverage elsewhere too?

4) Who conducted the poll?

5) What other polls have they done, and what kind of coverage did those polls get?

Other than that, just ask people. There’s brand awareness of the good pollsters, so even if you haven’t heard of any then someone around you probably will have.


So there’s potential for links in running a poll, but the following factors should be considered:

1) Subject Matter – Will the results be interesting to anyone? Who might link to them?

2) Credibility – Are there enough respondents, are the questions un-biased, and is a credible pollster conducting the poll?

3) Outreach – Do you have a plan to promote the poll results, which sites might be interested in publishing them, and how will you contact them?

4) Link Clean Up – How will you find sites that publish the results without attribution, and how will you go about asking for a link from those sites?

I’d love to know what you think. Have you run polls as part of your link building activity? Is there anything else people ought to think about and/or be aware of? Let me know via the comments 🙂