How user test­ing can im­prove your app


One of the most im­por­tant mind­set shifts that needs to oc­cur dur­ing scop­ing is un­der­stand­ing that nei­ther prod­uct own­ers or prod­uct de­sign­ers are the end user. Both prod­uct own­ers and prod­uct de­sign­ers nat­u­rally have in-depth knowl­edge about the ap­pli­ca­tion, but mak­ing the as­sump­tion that this knowl­edge is shared by the tar­get user can se­ri­ously af­fect that user’s ex­pe­ri­ence.

In ad­di­tion, whilst dur­ing scop­ing we want to dis­cuss our opin­ions and in­sight, it’s cru­cial to val­i­date all of these con­ver­sa­tion points with in­for­ma­tion from the end user. One of the con­se­quences of not user test­ing is sim­ply cre­at­ing a prod­uct that the tar­get mar­ket does­n’t re­spond to, be­cause the prod­uct in ques­tion does­n’t meet their needs. Often, this sit­u­a­tion arises be­cause the peo­ple most di­rectly in­volved in the app cre­ation process did not spend enough time get­ting to know the end user.

User in­ter­views

User in­ter­views are es­sen­tial for get­ting an un­der­stand­ing of the thoughts, feel­ings and mo­ti­va­tions of the tar­get user. For a suc­cess­ful in­ter­view, it’s use­ful to set a few goals about what you want to learn and how those learn­ings will af­fect the rest of scop­ing. At WorkingMouse we typ­i­cally scope against a prob­lem state­ment which pro­vides a con­cise sum­mary of the es­sen­tial is­sue to be re­solved by the app. This as­sists us in col­lect­ing data which is rel­e­vant to that prob­lem. User in­ter­views can be con­ducted with a panel of users sup­plied by the prod­uct owner (for ex­am­ple, the CEO of a com­pany al­low­ing some of their em­ploy­ees to an­swer ques­tions about an ex­ist­ing process), or a group of anony­mous in­ter­vie­wees who are called in.

When it comes to these anony­mous groups of users, there are two things that need to hap­pen be­fore­hand. First, the prod­uct owner and de­sign­ers must agree on the tar­get de­mo­graph­ics for the end user. Next, the method for call­ing in these anony­mous par­tic­i­pants must be de­vised. Generally, if this is the de­sired route, a third-party ser­vice which spe­cialises in this sort of user test­ing would need to be con­tacted so that they can vet the par­tic­i­pants.

User in­ter­views, con­trary to pop­u­lar be­lief, do not need large sam­ple sizes to be suc­cess­ful. Five in­ter­views per study is of­ten suf­fi­cient. The key thing to re­mem­ber is that user in­ter­views should not just be un­der­taken once. These stud­ies should re­peat of­ten with dif­fer­ent sam­ple groups. It’s not so much about the quan­tity of data ex­tracted at once, but the con­sis­tency of data gath­ered over time, as well as the analy­sis of the UX re­searchers.

User test­ing

User test­ing in­volves send­ing group of par­tic­i­pants ei­ther a pro­to­type or a func­tional ver­sion of the ap­pli­ca­tion. The par­tic­i­pants are then asked to com­plete a se­ries of tasks us­ing the pro­to­type or app. For ex­am­ple, if you were the owner of an e-com­merce web­site, you prob­a­bly would want to test how in­tu­itively your users could nav­i­gate search re­sults, add items to their cart and pro­ceed to check­out with­out any dis­trac­tions. User test­ing can be con­ducted in per­son, with the prod­uct de­signer act­ing as the in­vig­i­la­tor as the user car­ries out the tasks, or re­motely. In this lat­ter case, users will sim­ply run through the list of tasks pro­vided to them and pro­vide feed­back us­ing a form, sur­vey or sim­i­lar. It is of­ten quite use­ful to record the test so that the footage of a user jour­ney in ac­tion can be re­viewed at a later date.

As with user in­ter­views, a prod­uct owner may de­cide to test with a panel of cur­rent users or an anony­mous group that match the tar­get de­mo­graph­ics. Both sam­ple groups have their ad­van­tages and dis­ad­van­tages. Anonymous pan­els tend to pro­vide more ‘raw’ feed­back, be­cause these users might not have ever en­coun­tered the ap­pli­ca­tion or prod­uct pre­vi­ously. On the other hand, be­cause they are not cur­rent users or have not been iden­ti­fied as cur­rent or first users, they might not be as rep­re­sen­ta­tive of the tar­get mar­ket. If pos­si­ble, it is good to test with both groups to val­i­date as­sump­tions and iden­tify out­lier opin­ions. Unlike with user in­ter­views, user test­ing can be con­ducted with larger groups and be­cause they are less in­ten­sive than user in­ter­views, they are eas­ier to sched­ule in at any time.

Discover Software


Rhiannon Stevens

Get cu­rated con­tent on soft­ware de­vel­op­ment, straight to your in­box.

Your vi­sion,

our ex­per­tise

Book a con­sul­ta­tion