This week started with some interesting follow-up chats about ScaleSummit. I learnt a lot of things at ScaleSummit, but many of them were from colleagues at GDS. I'd like more of this kind of thing to happen within GDS, and as part of that I'm hoping to give an internal talk or seminar about building search systems, and in particular doing so using Elasticsearch. It's easy to forget how much knowledge we all have that isn't shared by other people we work with. This week started with our fortnightly "specialist" team show and tell, at which I showed off the search dashboard, which is working well. On Tuesday, I showed it off again at the GDS-wide show and tell, to an audience who hadn't already heard all about it while I was developing it. One (not entirely unexpected) piece of feedback from the content design perspective was that it would be great to get that kind of information about search performance related to an individual page into the content editing tools. I hope we'll be able to do this in the not-too-distant future. Tuesday was our sprint retrospective and planning day. There have been plenty of problems getting our new team structure going, but I'm glad to say many of them are quickly being solved; in particular, this week I spent well under an hour doing story planning, rather than losing whole days to it. We're also getting some momentum up, and moving stories into the "Done" column regularly. Our main problem now feels to be isolation, and I'm hoping that this will be improved by two changes. One is that we've all been working on separate tasks; there are 5 developers in the team, and we've been working on roughly 5 separate projects concurrently - one of these is now done, and we'll shortly have looser deadlines on the remaining ones, so we should be able to actually work on the same things for a bit soon. The other cause of isolation is a physical barrier which divides us from the rest of the office, and from the "program" team who are setting the longer-term direction of our work. We've been trying to get this wall removed since before I started at GDS, and there are signs this might well finally happen within a couple of weeks. As an experiment on Friday, I found a desk space in the afternoon on the other side of the wall, and had several useful unprompted conversations with people on the program team, which just hasn't been happening before now. Until the wall comes down, I'm going to try and sit on the other side of it whenever there's a bit of desk space free there, because these kinds of conversation are really important. I got a lot of time to sit down and work on the new API for rummager this week, and have a [pull request](https://github.com/alphagov/rummager/pull/204) open for this work. This should be reviewed and merged pretty soon, at which point lots of different search improvement tasks can start building on it - I'm very excited about this. I've also started writing notes for a blog post about the way we're using Google Analytics to measure search performance, which I intend to finish this week.