Yes. Arun, good morning. I appreciate the question and the pickup on the well performance. Obviously, really excited about what we are seeing there. And as we reported here today, we've now got 44 wells of the program online with production history and the performance is well above our 2023 type curve, which we are obviously enthused about. Really, the driver we are seeing here is the completion design piece that you mentioned. This has been a multiyear effort by our team to zero-in on the true causal factors that can drive well performance and increased recovery from shale. And it's not just in the Permian, we're deploying this across the whole portfolio. And really, I like to kind of break it down into three buckets, Arun, in terms of what the team has been driving on, and I probably won't do service to the detail and the multiple factors that are driving here, the three buckets we see really are the stage architecture, which is our phrase for the combination of cluster design, cluster spacing and then, of course, the sand and water loading that we are feeding in through those clusters. And the whole point of that stage architecture engineering is to get a more conductive fracture network more consistently placed along the entire lateral. And of course, that becomes more important as we drill longer laterals in these plays and drive efficiencies that way. So that's the first bucket. The second bucket is kind of what we call chemistry, and that's using the chemistry of the frac fluid try and enhance the permeability of oil in that fracture network. And so then, as a consequence, get more recovery from the poor system in place. And then the final bucket is really got a few things in it, but really, it's this idea of real-time frac monitoring. And the approach here is traditionally in our industry the approach was the engineers in the office would write up a frac program and send it out to the field and then the field would pump it. And then sometime down the road, there would be a look back on and potentially optimization occur in subsequent wells. And what we are trying to -- the network we are trying to crack with technology and telemetry is to be able to do that real time. And the learning from the signals we are getting back subsurface well we are pumping and then tune our frac design as we go and as a consequence, just be more efficient with that frac design, but also get more recovery along that long lateral. So really, those are the three buckets I'd speak to, Arun.