diff --git a/research/2025-05-14-Multiple_branches_concept_testing.md b/research/2025-05-14-Multiple_branches_concept_testing.md
new file mode 100644
index 000000000..f64131eb0
--- /dev/null
+++ b/research/2025-05-14-Multiple_branches_concept_testing.md
@@ -0,0 +1,77 @@
+# Multiple branches concept testing
+
+## 2025-05-14 / Sprint 17
+
+## Aims
+Learn about people’s mental models around multiple branches.
+
+Find out which elements of 3 concepts people find easiest to use to create multiple branches.
+
+**Design 1:** Works the same way routing currently does in the platform. Allows branching from any question or answer.
+
+**Design 2:** All questions shown on one page, where users can use dropdowns to select where form fillers go next. Allows users to add a skip from any question or answer.
+
+**Design 3:** Very different conceptually to current routing. Users add questions to a ‘branch’ from a specific answer in a radio list.
+
+## Participants
+- Civil servants. Existing users, with a range of experience.
+- Screened by survey to assess their level of experience of the platform
+- 8 (7 useable for analysis)
+
+## Methodology
+Concept testing of 3 designs.
+
+60 minute sessions.
+
+One task per design. Tasks carried out by the participant giving instructions to the researcher, who navigated screens on a Mural board, imitating interactions.
+
+## Key Headlines 
+According to the quantitative data, Design 2 was the easiest to use and the most preferred by participants.
+
+**Design 1:**  
+- Most managed to set at least some of the correct routes after figuring out how it worked.  
+- 4 set all the routes required    
+1 did not add the final route (not from a list question) and thought they had finished
+- 2 didn’t get as far as adding the final route  
+- ALL participants started by unnecessarily adding a route to the first answer in the list  
+- 5 people talked about how tricky it was to work out which routes to set where  
+
+**Design 2:**
+- They were able to add the skips easily.
+- Generally positive opinions and comparisons to the other designs
+- In particular having it all on one page was called out as helpful.
+- Having the ‘default route’ displayed while adding the routing made it easier for people to realise where they did and didn’t need to add routes.
+- All but one participant struggled with the text of the next question not being shown.
+
+**Design 3:**
+- Mostly they manage to add the correct branches for the two answers that needed them.
+- Quite often there was initial confusion but they worked it out in the end.
+- 4 people thought they also need a route for the answer that went straight down the ‘default’ route.
+- There was quite a lot of thought put into what a branch actually was and how it worked
+- 4 people mentioned they liked the process of adding all the questions at once into the branch.
+- There were a lot of queries and mistakes around the mechanisms of adding the questions. Most expected to add the detail of each question before adding the next question.
+
+Quite a lot of conflicting feedback and responses to using this design:  
+- Some were unsure where the branch ended or how it connected back to the main form. Most did eventually work it out or make correct assumptions.
+- Some said it was less intuitive, while others found it more intuitive.
+- Some were confused, while other found it quite easy to understand.
+- The more experienced users found this concept easier to use.
+
+**Broader observations**
+- People noticed whether a design worked in a similar way to current routing (adding questions before adding routing) or differently. Some found it helpful when it worked the same way as currently, but others said they liked adding questions when adding routing instead of before.
+- It can be tricky for people to understand that form fillers will go a ‘default’ route. This is seen in Designs 1 and 3, where people build a route/branch.
+- As a result, many want to add a route for all branches. Only experience or a lot of thinking helps them realise why they don’t need to. In Design 2, because it’s visibly demonstrated as being in place (‘go to q2’), people understand more easily.
+
+**Conclusions**
+- Design 1 is most challenging for people to use.
+
+- Design 2 is easiest to use and preferred by more users.
+
+- Design 3 had challenging aspects, but was well liked by some and does have positive aspects, particularly for more experienced users.
+
+- Designs 2 and 3 both have potential opportunities and risks that should be explored, whilst keeping in mind that Design 2 was by far the easiest to use when considering all experience levels of participants.
+
+## Supporting Evidence
+- [Report](https://docs.google.com/presentation/d/1wO49c1MD9HzH_VLVHu4nN_BqiRb2j9fnnzgmh_MddrI/edit?slide=id.g3619d51f40a_0_67#slide=id.g3619d51f40a_0_67)
+- [Playback (sprint 17 demo)](https://docs.google.com/presentation/d/1wO49c1MD9HzH_VLVHu4nN_BqiRb2j9fnnzgmh_MddrI/edit?slide=id.g3619d51f40a_0_67#slide=id.g3619d51f40a_0_67)
+- [Further documentation](https://drive.google.com/drive/folders/17tSiooPfMI9GpGTCKdGqNKySkjV-Rwvi)