A Framework For Creating User Personas, Fast

User personas (or buyer personas) are too often associated with presumption and guesswork. However, there are highly-analytical ways of establishing sturdy profiles that are backed by research and data. CXL Institute provides training and insights into conversion rate optimisation, a key part of which is a data-driven approach to user personas. This framework is sourced from their methodology.

First…what are user personas?

User personas are semi-fictional representations of your ideal user or customers, a model of what they want to achieve and why, as well as other relevant characteristics. They are presented as a profile that is used internally within an organisation to help guide marketing and product development decisions according to these established wants and needs. As you can imagine, this is a pretty big deal, and not something that should be neglected.

However, they are often poorly made, or simply erroneous altogether, which can misguide your wider business efforts.

Some irrelevant information that is typically included in these profiles are favourite colour, favourite food, etc. These characteristics can be considered to be superfluous, as they don’t have any bearing on the things the user is looking to achieve (with rare exceptions, of course).

A rigorous framework for creating realistic, reliable user personas

These portraits are often, indeed, pulled almost out of thin air, based on assumptions and one person’s — or few people’s — understanding of what the ideal customer looks like. But this is just what they want them to look like, and the reality can be slightly or entirely different. With a data-first, analytical framework, you will be able to create a reliable and realistic model of what your average users look like.

The framework consists of 3 steps, and can be completed, start to finish in 2–4 weeks:

We can go through each of these steps in context: let’s say we’re working on building a platform to help people buy cars.

Step 1: Collect quantitative data

What’s the best way to build out hypothetical personas that represent your ideal customer? By talking to those that are interested in your industry, people that would form some part of your target market, and asking them questions that are:

  • relevant;
  • actionable; and
  • unbiased.

You should develop a survey that is carefully devised as to draw actionable, qualitative information that can be presented quantitatively.

A way this can be achieved by presenting multiple choice questions with descriptive scales instead of numbering them in order of intensity. Let’s take our example of working on a platform to help people buy cars. An example question might look like the following:

Where some surveys might ask the respondent to select a number from 1 to 5, this method provides more specificity. This helps to eliminate noise in your data, and no respondent can confuse the extremities of the scale, or divide the range differently. Importantly, however, we will treat these qualitative answers in a quantitative fashion when we later process our data.

When it comes to setting up the survey online, there are a multitude of tools available to host on:

The tool itself is not important, as long as it lets you ask open-ended questions. This lets you take quotes from interesting answers which will give you invaluable “sticky copy”.

PRO TIP: To verify legitimate respondents and weed out anyone illegitimately filling out the survey, include a random, unrelated, easy-to-answer question. This will let you know who isn’t paying attention or closely reading the questions, and you can exclude these respondents’ answers from the final data set.

Ideally, you want to source anywhere from 300 to 1000 respondents for your survey.

“Uh…where am I going to find that many people, much less relevant ones?”

Great question! There are a couple of trusted methods. If you’ve got enough traffic, you can literally invite visitors from your site — tactfully — to come in and take a survey using a tool like Hotjar. If you don’t have decent enough volume to meet that sort of quantity in a decent time frame, try Amazon Mechanical Turk; for a small price, you can be given access to the people you’re looking for.

Great, let’s dive into surveying!

Hold up. Before you spend any significant amount of your budget or time, be sure to conduct a pilot test. Test the survey on 10 to 20 people, go through the whole analysis process outlined ahead, and share the data with your colleagues to discuss the results.

Step 2: Statistical cluttering

After a few days, you should have hundreds of survey answers. Now, you can look at organising the information into tidy data. This is defined as:

A standard way of mapping the meaning of a dataset to its structure. A dataset is messy or tidy depending on how rows, columns and tables are matched up with observations, variables and types. In tidy data:

1. Every column is a variable.

2. Every row is an observation.

3. Every cell is a single value.

From the tidy data we can start to look at doing an Exploratory Factor Analysis (EFA). This technique will help you reduce all this information into smaller subsets of data. These will become your “factors”, which represent the common underlying trends in survey responses. Starting to see the personas form?

If you posed our questions correctly (remember “relevant, actionable, unbiased”?) then this shouldn’t be difficult. To execute on this, there are a wide range of tools at your disposal.

You can specify how many factors you want to sort into, and decide on the weighting each factor has. From here, “clumps” will begin to form in the data based on underlying commonalities.

Remember to visualise the data with whichever tool you have opted for in your factor and cluster identification — this makes it much more presentable and easier for you to identify data integrity issues.

Step 3: Building archetypes

Finally, you can translate your findings back to plain, readable English for use across your organisation. In doing this, you should stick to a few key guidelines:

  • Always start from the user’s needs
  • Don’t force “fluff” info (favourite colors, most demographic info, etc.)
  • Include at least one quote for each persona — these should be real, “sticky” quotes provided by survey respondents

Consider collaborating internally with product managers and designers to build out a user journey timeline, plotting what a typical sales cycle looks like from their perspective. Also throw in a list of resources, tools, and websites that each user probably find useful along their journey.

Once you have these personas done up and distributed internally, make sure you don’t get complacent in the long term; these generally expire within 1 year based on evolving value propositions and objectives within a business. So, at least yearly, look at doing a user persona review.

Until that point, these are your beacons of truth for your marketing messaging and product design. Get to the point where you always have them in mind, however, remember to validate them along the way.

Trust but verify.

--

--

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store