 By now, you should have an overview of all the parameters that can impact noise. So you know that there are a number of parameters involved, and on top of that, the way these parameters interact with each other can be a little bit complex as well. So this can be quite difficult to reason about. So in order to understand noise, what would be helpful is to be able to pick a few parameters, see how impactful noise is with these parameters, then change these parameters and see how noise ratios have changed. And what I've really just described here is a simulation process. And well, this is exactly what NozLab is. NozLab is a simple simulation tool. Here's what your workflow with NozLab may look like. Step one, you would input some data and parameters, and by data I mean dummy data, data that would look roughly like real measurement data. Step two, you would run a simulation, which means that you would ask NozLab to generate summary reports based on your data and your parameters. That would look roughly like summary reports you may get from the aggregation service. Step three, you would observe the noise ratio for these summary reports. Now, what's really important to note is that in reality, if you're using the API and the real aggregation service, you don't know how much noise you have. But with NozLab, because we're simulating, we're actually giving you a peek under the hood and we're letting you know what your noise ratios would look like. With this data and these parameters, so for your assimilation. So with this, you've completed a first situation. And let's say in this example that the noise ratios look a little bit high for your use case. So you may want to repeat the whole process. So then you would tweak the parameters, run another simulation to see how those ratios have changed with your new parameters and so on. With this workflow, you can see that NozLab can help you achieve two things. The first one is to build up your understanding of the parameters that impact noise and sort of develop an intuition for these parameters. And the second thing is to quickly test out strategies to reduce noise ratios. And you can also use NozLab as a way to demonstrate these strategies quickly to others in your organization. Now, this is quite handy, but I would like to call out a few strengths and also limitations of NozLab just so you can make the best use of it. First, a few strengths. NozLab is modeled based on the real API. What this means is that the simulation uses the same mathematics, the aggregate service uses to add noise. So for a given set of measurement data and parameters, the noise analysis you get in NozLab should be similar to the results you would get with the real API. Second, NozLab doesn't require any technical setup. You really just need to open your browser. And it also gives you access to kind of a direct feedback loop where you can access some in-line help and in-line definition for each of the parameters and then you can tweak those parameters and see if the noise analysis you get as a result actually matches your understanding of that parameter. So much for the strength of NozLab. Now, I would like to call out a few of its limitations. So first of all, NozLab doesn't capture to the full extent of possible ad-tech distributions and scenario. It's a simulation tool. And one example of these limitations is that we have limited flexibility on the input measurement data. A second important aspect is NozLab is not an end-to-end simulation tool. It does not integrate with ad-tech systems, but it only really captures the API functionality. Now, once testers have used NozLab to onboard and are comfortable with key noise concepts, at some point they're going to reach a stage where they are going to need fine-tuning and a lot more flexibility in simulating their measurement data. And in that case, they can switch over to another tool we've made available for them and that's the Simulation Library, which is more powerful, but also requires some engineering know-how. So the Simulation Library is out of scope for this presentation, but I wanted to bring it up just to give you a glimpse of the tooling landscape here. Back to NozLab, a few other things you should know. NozLab is completely open-source. You can check out the source code over here on GitHub on that link. The tool is also experimental. The code and algorithms have been verified, but this is important to keep in mind as you're using the tool. And we've really evolved NozLab based on tester feedback. So we're always very much open to feedback on NozLab and how we could enhance the tool to make it as useful as possible for testers. If you find an issue with NozLab or if you have an idea or a feature request, we'd be happy to know about that, so please share with us. And one last note, we have a companion doc or user guide for NozLab, which is publicly available on our developer website. You can check it out at this URL.