As I mention in [my answer][1] to the question on [What research methods can I use to create personas?][2], some key elements to keeping the persona valid and relevant include: - **state your research goals** before you start so that the persona can be kept relevant to your goals - **define how you will use personas** so that you can ensure your personas include the relevant detail and content to move directly to the next phase of the process - **interview real users** where possible - **undertake ethnographic research** where possible to get context for your content - **review as you go** so that you can assess the persona against your goals, and are working towards the achievements you set out to meet, and be prepared to adapt if necessary - **keep it real** by always following up on information that may be second hand research such as that from proto personas - **avoid [bullshit personas][3]** at all costs - **make it collaborative** so that everyone has input, and gets a chance to have their say. If everyone takes the journey together, then everyone can see the information has a history and a source, and can be used as evidence, rather than some black box of information gathered by an outside agency for example. **Just keep the bs out of the persona** and remember - **Personas are not just an artifact** See the above linked answer for more detail on each item above and a handful of excellent references on personas. Edit: **So this helps you get good quality personas** - a kind of self validation enforced by the process itself. But lets say you have your persona and you still want to increase the level of confidence in the content - post production validation - proving the personas are not just an artifact after all - proving it's not just bs. Here's some ideas: - **Get wider domain expert level approval** - You may have worked with a domain expert in the the creation of the persona. But the people involved in producing the personas should encompass different backgrounds with different expertise and with different types of input. So once you have your persona - throw it a *whole bunch* of domain experts and see what feedback comes back. - **Complexity breeds inaccuracy** - Personas are by their very nature rather vague and inaccurate - they are a *representation*. So the deeper the level of information and the more accurate the level of data contained within, the greater chance of either the persona being mistaken as *fact* rather than *guide*, and the more likely the information is to just be wrong. So keep the information simple. Too much information - the greater chance of it being bs. The less specific the data the less tightly coupled the rest of the process that depends on the personas. - **Validate with quantitative data** - Coupled with the previous item - if there is a representation of accurate quantitative information in the persona - post validate the persona against the *facts* you collected during the process. Question everything. - **Seek *self* identification** - Interview users again afterwards. Ensure that some are the same and some different. Get them to identify with one of the personas you have created. If nobody identifies with any the personas, something went badly wrong. Bear in mind that you will usually get *someone* who cannot match themselves with one of the personas. - **Seek *observed user* identification** - Interview those connected to the users or customers afterwards. Get them to identify the people they deal with from amongst the personas. Again, if people who are so tightly linked to the target audience cannot identify them, something went wrong. - **Gauge persona relevance** - Review surveys and answers to questions that were raised during the creation of the personas. Match personas against the answers received. If only a small number of personas take the bulk of the matches then the information was not evenly distributed and the less relevant ones should be reconsidered. Understand however, that even if validation appears to have gone well, it's not until later in the process when your product or service is put in front of real users for usability testing, and is released in a real marketplace, that the true validity of the personas can be confirmed - *or doubted*! [1]: https://ux.stackexchange.com/a/21916/6046 [2]: https://ux.stackexchange.com/questions/21891/what-research-methods-can-i-use-to-create-personas [3]: http://avoidbspersonas.com/ [4]: http://www.thinkui.co.uk/2012/06/what-research-methods-can-one-use-to-create-personas/