Website Personas: A Practitioner’s Guide – The Research (3 of 7)
Personas are as strong or as brittle as the research from which they emerge. As the graphic shows, we utilize, when we can, five categories of data. You may have neither the time nor the budget to explore them all. The point is simply that “eureka!” insights can emerge from anywhere, so it is worth investigating as many data sources as you can efficiently sift.
- Existing research: Our clients tend to be Fortune 500 companies, which typically means there is an abundance of existing, primary research. The trick in these situations is parsing out what it relevant to the task at hand. This type of data is useful for establishing base demographics, as well as market segments and their defining characteristics. (Market segments can overlap with personas, but they can also be radically different. Personas are best derived from need state analysis – more of which we will discuss below – rather than market segmentation.)
- Channel-derived user experience (UX) data: Channel data can come in a variety of forms. Web analytics is an abundant source of behavioral data. Call-center or customer service reports and transcriptions are also excellent sources of anecdotal data, as well as behavioral or attitudinal trends. Every touch point will have data of some relevance to the persona creation process.
- Site survey data: The third box in our graphic depicts survey data. In the interests of full disclosure, I should point out that this is where our proprietary survey technology gives us a competitive advantage over other persona developers. Our WebIQ technology captures survey responses and a respondent’s clickstream behavior as an integrated record. We know where a visitor travels within a site, along with what they do (and what they do not do). So we can ask questions of them based on content they accessed, tools they used, or behaviors they exhibited. This gives us direct UX metrics for every aspect of an existing site, along with direct user feedback about specific site content and functions. WebIQ data is rich in the attitudinal, behavioral, and experiential insights that endow our personas with authenticity and do so much to bring them to life. That said, survey data gleaned through regular HTML surveys is adequate if behavior-based survey data is not available.
- Qualitative research: This can take the form of ethnographic research, in-depth interviews (IDI’s), or focus groups. One of our clients tells of a persona initiative she worked on for an accounting and tax preparation firm. She describes the culture as “one of numbers,” by way of explaining why the company conducted 5,000 IDI’s and undertook months of extensive ethnographic work and dozens of focus group sessions to secure what they felt was statistically sound qualitative data. This type of rigor is viable only if you have unlimited budget and resources, along with an open-ended timeline. Rarely is that the case, however, so the typical approach is to conduct as many interviews or run as many focus groups or undertake as much ethnographic work as time, resources, and budgets allow. The qualitative research is best undertaken after the primary quantitative research, survey data, and other data sources have been analyzed. The findings from these quantitative sources will inform the question sets for the qualitative research. A key word to listen for in the qualitative research (as well as when scouring survey comments) is “that”, when used as a relative pronoun. When someone starts with the phrase “I think that …” you are about to hear a belief statement. Capture it. It is pay dirt.
- Secondary research: This, typically, is undervalued. It should not be. Secondary research, by which we mean any source of information not contained in the previous four research categories, can add industry or social context from perspectives far removed from a brand or organization’s internal take on the world, their markets, and their customers. Analyst reports, industry reports by trade groups or private equity firms, newspaper articles, blogs, Facebook pages or Twitter feeds, Better Business Bureau data, product reviews, or community sites can all be viable and valuable sources of detail and context. Time often precludes wide, deep perusal of secondary research, but it should be mined as rigorously as possible. It is well worth the effort. I always Google “day in the life” + “architect” (or whatever the profession or profile of the target audience). The results invariably turn up nuggets.
Different “need states” take shape out of the persona research. The extent to which needs differ from user to user helps in identifying different personas, since they should inhabit, as much as possible, distinct need states. The subsequent need state analysis lays out the skeletal structure around which each persona takes shape and substance.
As with almost every aspect of the persona development process, there are a variety of ways you can approach need state analysis. If you have the time and expertise, regression analysis is the most scientific and defensible method.
If you are not a statistician or you do not have access to a statistician, you can build a matrix with the needs listed on the vertical axis of a spreadsheet and add personas column by column to the right as different need states emerge from working down the vertical axis. The operating principle driving this method says that if you take care of the needs of Persona A, you will have also addressed those particular needs for all personas. You then move to Persona B, whose needs, for the most part, will be different from those of Persona A. From Persona B, you move to Persona C, whose needs are, by and large, different again from those addressed through Personas A and B. Need state analysis conducted in this way also allows you to decide if you need to divide your persona set into primary and secondary personas.
The spreadsheet analysis can then be consolidated into a more readily consumed deliverable. I once saw a particularly elegant need state deliverable built by Carlson Marketing.
It makes for a very efficient exercise if you can reduce the analysis to two axes. Often that is not possible, but very powerful when it is.
Whatever your chosen method, spend time on your need state analysis. It is the final “sign-off” document before you start building out your personas. As with any building process, mistakes made in the blueprint will cost more and more to rectify the later in the process they are discovered. Devote as much time to the need state analysis as it requires to get it right.
– Roger Beynon, CSO, Usability Sciences
Interested in talking with us about Personas? Contact us over at getdirection.com