Last week I paid a ton of taxes, and it filled my heart with joy. I was so happy that I felt like jumping in the air. Did I lose it completely, to pay so much money, on taxes, no less, and be so happy? I’ll let you judge for yourself by sharing what thrilled me. 

I’ll start by saying that it wasn’t because paying a lot of taxes means you run a successful business. It probably does, but the taxes I have to pay are the taxes I have to pay, and in and of themselves there’s nothing to be happy about. It is, however, related. To explain how I’ll need to give you some background (I’ll keep it brief!) on how the Israeli tax system works. 

When you own a business in Israel, you pay income taxes every month or two, as a fixed percentage of your revenue. This fixed percentage is set by the tax authorities based on what they know about you, which is your revenue and how much taxes you paid last year or the year before that. 

When I started my business, I hoped that the tax authorities would set my rate to be fairly low. And when they did, I was happy to pay as little as possible. But the problem is that the real rate that I needed to pay would only be calculated when the year ends, taking into account not only the revenue I made but also the expenses I had because income tax is paid on profit, not revenue, and that calculation is done as part of your annual report. Since the annual report takes time to prepare and file and is typically filed in Q4 of the following year, I wouldn’t know the real amount of tax that I would have to pay until then. To make it worse, since my business is growing, the rate that was set for me by the tax authorities doesn’t represent the real rate that I would need to pay eventually, so I know I would need to pay them more money. How much more? I wouldn’t know until Q4 of next year when I file this year’s report.

Yes, you got it right. It means that I owe the tax authorities what could be a lot of money, but I wouldn’t know how much for almost two years. How can one make any business decisions like that? Whenever I looked at the balance on my bank account and wanted to calculate my next steps, I couldn’t, simply because I didn’t know how much of that money really is mine.

What happened last week is that I sat with my accountant, we estimated a close approximation of the real tax rate I would need to pay, and not only did we decide to pay the adjusted rate from now on, I calculated the tax debt I already had out of paying the rate that was originally set for me for this year, and paid it fully, in advance. The result was that I knew that the next time I would look at my bank account, I will get a much clearer picture of reality – a crucial precondition to making any informed decisions. The reason it made me so happy was that I knew that now I can make progress. I wasn’t in the dark anymore, despite having significantly less money in the bank I felt the weight lifted off my chest, almost physically.

If you want to make data-informed decisions, here are a few ground rules that you must follow.

Make Sure Your Data Is Accurate

This first guideline sounds simple, but it often isn’t so: if you want to trust data as input to your decision-making, you must make sure the data is accurate and represents what you think it represents. The first and most obvious step here is to clean data that simply doesn’t need to be there, for example, product activity that comes from internal users and not from real customers. But that’s just the first step. 

In many cases, there would be hidden biases in the data. For example, when I worked at eBay, and we wanted to understand how well we are doing across all of the inventory eBay had at the time, there was a single category that was so huge that any trend there impacted the numbers on all of the other categories, even though the trend was local to that category only. To overcome this, we calculated most of our metrics without this category, which had its own set of metrics. Another example is numbers that balance each other, for example, growth in active users that covers for a huge churn rate (meaning you are acquiring many new users but at the same time are losing users as well). In these cases, the solution is usually to add multiple sets of metrics to make sure you truly understand the world with the data you have at hand. Remember that the world is multidimensional, and there isn’t a single number that can accurately represent it.

Understand What You Are Looking At

As in the last example above, many times the numbers seem nice and clear, but when you start asking questions you realize that you don’t really know the answers, and as a result, you don’t fully understand the data that you are looking at. The problem with this is that numbers without full context can mean very different things than what you think they mean, and would cause you to make very different decisions. This is one of the insights that the CPO Bootcamp participants get as part of the data-driven product leadership module.

One way to know if you fully understand the numbers is to explain them to someone else. It can be a colleague, a team member, a friend, or in some cases your manager. It would be best if it is someone who knows their way around numbers and would ask you hard questions that would help you make sure there are no gaps in your understanding. Domain expertise plays a role here too: if you are looking at your SaaS metrics, someone who has worked with such metrics before would be able to ask more concrete questions than someone who only worked in eCommerce, for example.

You can also put yourself in the reviewer’s chair, and ask your team members to present their analysis of the data. This allows you to think about it objectively and keep an external perspective. Another way to make sure your data and conclusions are intact is to ask yourself where can you be mistaken. When you have the conclusions, ask yourself what in the data could have been misinterpreted, that would lead you to the wrong conclusions. For example, if you realize that churn is not a problem, ask yourself what additional data could make you change your mind. It might be something like ‘churn, in general, is not a problem, but churn in a specific segment is actually alarming’. Note that you won’t necessarily have access to all of this data upfront, but by asking the right questions you might realize you need to ask for this additional data.

Know When to Stop

Being fully data-driven can be a never-ending story. Attempting to get to perfect data might hurt you in two ways: the first and most obvious one is that it might be impossible or way too expensive to get perfect data. If you attempt to get there, at some point the effort wouldn’t be worth it. But it’s the second one that is more dangerous: since you will rarely if at all get to perfect data, you must know that you are working with incomplete or imperfect data, and manage the risk accordingly. If you think you are making a decision with full and accurate data when it is usually not the case, you are taking a much bigger risk than you intend to.

Understanding that data would never be perfect, it is important to know when to stop perfecting it. There is a sweet spot where the ROI is diminishing, or when getting more data wouldn’t make a huge difference.

For example, in my case above, the estimated tax rate that I got to with my accountant isn’t 100% accurate, but it is a good enough approximation to help me with the actual decisions I needed to make. Had I strived for fully accurate data I would have probably been stuck in analysis paralysis (or endless expense reports). On the other hand, I know that this is not 100% accurate and I take it into consideration when making my decisions.

Data is such an important tool for your decision-making, that you must make sure it is useful for you. Some of its usefulness resides in its availability and timeliness, not just in its depth and accuracy, so make sure to always balance these two dimensions.