Press "Enter" to skip to content

The Problem with Personas in Digital Marketing

Personas are a marketing concept that has been around for decades. The idea of using personas probably originated from the hallowed Ogilvy agency, and has since gathered many fans in marketing circles.

In recent years personas have also become a popular approach in digital marketing. Early mentions of using personas in online marketing go back as far as 2007, and in recent times have increased exponentially as more internet marketers explore the possibilities and blog about it.

There are many professed uses of personas, from application to UX-design to keyword-level personalisation. And there are as many vocal advocates of personas as there are use cases.

Using personas


But there’s a problem with personas. And that problem is that defining personas is mostly guesswork.

Once you have a persona it’s easy and fun to imagine how this fictitious person would use your website, search for your products, or engage with your service. But the persona itself is the product of guesses.

At best, personas are extrapolations of a modicum of demographical data – from consumer surveys, social media analytics, or general market research. At worst, they’re purely wishful thinking lacking any foundation in reality at all.

Even the more data-driven personas are derived from idealistic notions of what a potential customer looks like. They often end up as cardboard clich̩s, two-dimensional characterisations that Рat most Рreflect only a tiny proportion of actual users.

I can see the value of using personas in scenarios where decisions have to be made on the direction of a certain marketing message where there’s no data to rely on.

But in digital marketing, I believe personas are a waste of time. There’s a much more effective way of achieving results without resorting to guesswork, hopeful extrapolations, and wishful thinking. That way is, of course, through testing.

Data-Driven Decision Making

On the internet we have the luxury that we can present random selections of users with different versions of our sites. We can do A/B split-testing or multivariate testing to determine exactly which design, what content, which call to action, works best.

We don’t need to invent fictional personas to try and decide on a specific type of marketing message. We can simply create multiple messages and test which one yields the greatest rewards.

We don’t need to think of theoretical keyword scenarios to personalise landing pages based on what we believe a user might like to see. We can simply extract the keyword from the referral string and chuck it in the landing page, and split-test the remaining (not provided) traffic.

There’s quite simply no need for us to rely on the horrendously flawed personas methodology, because we have much more effective methods at our fingertips. Instead of guesswork, we can use actual data.

And yes, the tests themselves need to get input from somewhere, to inform what to test and what a variation should look like. But this too doesn’t need personas. It just needs data from existing sources (again, social media analytics like Facebook Insights come in useful here), and a healthy dose of creativity and a “˜let’s try this‘ mentality.

I genuinely can’t think of any online scenario where personas are more useful than data-driven insights. I’d love to be proven wrong, though.

I’m keen to hear what you think. Do you believe there are cases where the use of personas online is preferable to testing? Please leave your thoughts in the comments.



  1. jpandian July 2, 2012

    I disagree, if you define personas the way most marketing folks do then yes, its mostly guesswork.

    But what if you define personas through the keyword data of a website, isn’t it more accurate then?

    Don’t we have to understand who’s visiting the website in order to be able to sell anything to them?

    Isn’t it more difficult to sell to someone through a/b or multivariate testing if they are just a formless mass without an identity?

  2. Jason Stearns July 2, 2012

    Personas are still necessary as they help businesses make a variety of business decisions, not just web pages. While they may start out as guesswork, personas are meant to be refined over time.

    I can’t imagine a business relying on social media data. It’s far too unreliable for any business decisions. Nor do I think taking the keyword from the referral string and chucking it into landing page will do the trick either. Thought needs to go into the content creation and that is done partly through personas.

    Using actual data is great and should be done, but personas still have a place.

  3. Barry July 2, 2012

    @Jey: using keywords to extrapolate in to personas is dangerous, imho, because it risks making seriously flawed conclusions based on horrendously incomplete data. There’s almost nothing sensible you can say about a person based purely on the keyword – all you can meaningfully say is probable search intent. And that’s not a persona by any stretch. Anything beyond search intent is pure guesswork.

    @Jason: making business decisions based on personas is deeply flawed, because those personas are not rooted in any kind of reality at all. It’s all just wishful thinking and baseless assumptions. That’s not a good way to conduct business. Testing is. Let data feed the decision making process. Show me how using personas is superior to data-driven decision making and I’ll change my mind.

  4. Richard Fergie July 2, 2012

    Are personas ever preferable to testing? No
    Are personas completely useless? No

    I would say that when you decide what to test next using whatever data source you are actually creating a persona in your head and anticipating how that person will react to the change. If you don’t think of this as using a persona then my argument doesn’t have a leg to stand on.

    Going through a formal persona creation process can help you improve communication in this case because you can refer to something/someone that everyone understands.

    So if you present personas as an alternative to data driven decision making, I agree with you because nothing beats data. I prefer to think of them as a complement to the data to make it easier to communicate concepts and ideas.

    Whether or not it is worth creating personas given the risk of whoever is in charge using them at the expense of data is another question 😀

  5. Barry July 2, 2012

    @Richard: two points here: I would argue that using data sources to predict possible click behaviour is not really using a persona – you don’t need to create a fictional name, job, background story, etc, and thus you’re not really building a persona.

    Second, one thing that happens pretty much every single time when you start doing A/B tests is that the outcomes are counter-intuitive. What you think will work, doesn’t. What you think will bomb, instead boosts.

    That too is why I think using personas is flawed, because what we expect people to do (fictional or real) very often runs counter to what they actually end up doing.

  6. Richard Fergie July 2, 2012

    @Barry: Your second point covers using personas over data so I think I am in 100% agreement with you there.

    When you predict possible click behaviour you are building a crude mental model of a person. Giving the person a name, job, family etc. is unnecessary because it won’t change the value of your prediction. I would still call this a persona because it is a model of a person.

    I think it is your final point that blows the biggest hole in my arguments: How do you update a persona when the predictions derived from it turn out to be false?

  7. dan barker July 2, 2012

    hi, Barry,

    I like the provocative, angry blog post you’ve written here 🙂

    Saying “data is better than personas” or “testing is better than personas” doesn’t make sense, as there’s no need for them to be mutually exclusive. If anything they’re probably more valuable combined than they are apart. (though yes – if you had to pick one – it would be good data every time).

    Personas can be useful. Sometimes they’re useful simply as a fast, memorable communication device for a bunch of data points on actual/ideal customers. Sometimes they’re useful to settle simple niggly arguments & save time. Sometimes they’re useful to retain focus over time. And, sometimes, they’re not useful at all.

    I think the issue you’re actually talking about here may be less to do with the concept of personas, and more to do with ‘cargo culting’. ie, people come across a concept and mimic the outward appearance of it before they fully understand the purpose, the nuance, how to use it in the most useful way, or the most appropriate place to use it.


    ps. “Early mentions of using personas in online marketing go back as far as 2007” – I bet you £100 I can find a mention years earlier than that 🙂

  8. Barry July 3, 2012

    @Dan: I don’t deny that personas have uses in scenarios where there is no data to rely on. My point is that in digital marketing, such scenarios are almost non-existent and that the energy devoted to developing personas – with all the flaws I’ve outlined above – is better spent on developing tests that inform decision-making.

    Also, I won’t be taking that bet as I’m pretty sure you’d win. 🙂

  9. dan barker July 3, 2012

    hi, Barry, here’s a devil’s advocate scenario for you:

    I work for an independent toy retailer. We mostly sell craft toys. The boss has decided to launch 2 new ranges:

    1. A monthly craft hamper.
    2. A brand of french toys – the biggest in the market there – who don’t have any significant distribution in the UK.

    I need to instruct a new agency to put together an email program around each of these, a copywriter to write product copy, catalogue copy, and email copy, as well as more complex landing page copy to explain the ‘craft hamper’, and I need to speak to insert agencies about where best to place inserts to drive new people ‘like my existing customers’ to the craft hamper landing page.

    There is piles of data around my customers; there is also plenty of data around the french market; I have data on previous landing page copy tests I’ve done, and email copy tests; there is a ton of data on people who buy ‘monthly recurring billing’ products; and there is lots of data on both the craft market & the branded toy market.

    Each of these third parties would like to know who my ‘customer’ is.

    1. Which data do I hand to them to explain my customer?
    2. How much is it going to cost me for them to spend time going through that data?
    3. What’s the likelihood they’ll be able to correctly interpret that data?
    4. They each understand the well-worn concept of personas, and I’ve got some pretty decent ‘rule of thumb’ personas distilled from my own data. Should I deliberately not show them those, in favour of raw data?


  10. Digital Agency July 5, 2012

    In marketing and user-centered design, personas are fictional characters created to represent the different user types within a targeted demographic, attitude and/or behavior set that might use a site, brand or product in a similar way. Use of personas in defining a marketing strategy is not new. Among others conversion optimization practitioners, web developers, content strategists and usability experts have been using personas for ages.

  11. Arjun K M July 11, 2012

    I definitely agree Barry,
    For more insights, there is a webinar which is being hosted by Infosys BrandEdge on July 18, 2012 about ‘Accelerating Global Digital Marketing’
    You can follow the link below to register

  12. Ed September 7, 2012

    I think this article is misguided. I dont think you have ever developed a persona in your life otherwise you would never say it is guesswork.
    I have been part of a digital marketing team for many years and we have used personas to develop strategies in search, design etc very succesfully. It is not guesswork, as you point out yourself there is a lot of creativity involved but it is all based on data and insights. We certainly don’t sit there dreaming up pretend people without any foundations that you erroneously point out.

    I totally disagree with your notion that you somehow cant do any split tests if you go down the persona route! You need data to do the split tests in the first place and in my opinion there is no better start than by developing personas. We are not naive to think that the personas we have developed are the finished product – they are just the start of a process that involves, keyword development, testing and refining. What we end up with are solid accurate representations of our target audiences that can be used in full service agencies to develop, websites, campaigns and other strategies.

    Admittedly you need a decent budget and consequently not everyone can develop them as they can pay for the variety of tools that assist the development of a persona. It all which takes time, and to the the cheap rankings obsessed SEOs this is not economically viable.

    The reason why I like personas is because it is the very essence of marketing. Marketing is all about the customer and nothing defines the customer better than personas.

    I think you are getting bogged down with the 1950s approach to personas i.e. deciding on a name, job etc. This is not how we develop personas, I dont care what the names are of a persona, I’m interested in their pain points, why would they go for one product over another? are they more likely to buy over someone in a different demographic, why? How are they solving their problems currently? how can we make their lives better? and so on… and not a name in sight.

    At the end of the day we are happy and succesful with our approach over guessing which keywords somebody might use in a search engine and working backwards or worse asking the client to give you a list of their favourite keywords.


Comments are closed.

Copyright© 2010-2024 Search News Central (SNC) | No material on this site may be used or repurposed in any fashion without prior written permission.

Search News Central uses Accessibility Checker to monitor our website's accessibility.