Speaker 1: Let's have a look how you can actually use ChatGPT for market research. We will concentrate only on looking at the competitors. Obviously you can do other parts of the market research as well. We will look at the furniture market in Poland because I have a quite good understanding of this market. At some point in my life I had to do a market research on that. I will go through the things that he has generated during my chat with him and I'm gonna explain you what kind of problems may occur. So first we started by asking him to act as a management consultant. Usually this helps him to think in a specific way. After that I asked him to list the main producers in Poland. I got those 10 producers. As you can see some of them are here twice. So black, red, white, section number 1 and 10. After that I asked him to estimate the revenues and we got it for 4 of the players in Polish zloty which is local currency and US dollars. Since we wanted to have it in a more structured way we asked him to put the data into 3 columns. So name, revenues in Polish currency and US dollars. And obviously we asked him to bring back the players for which he did not have the estimation. So at the end we got something like that. So a table with 8 players, for 4 of them he had revenue estimation, for the rest he did not have data. We didn't like the way he put the data so some part of data like Vox are in millions, the others are in billions. So I asked him to put everything in millions and moved the name of the unit to the name of the column and this is what he actually did. So we got the revenues for all the players in millions, both in Polish currency and US dollars. Now the next piece of information we wanted to have is the market share. So first I asked him how big the market is and he got some estimation that it should be around 10 billion, at least it was in 2020. And based on that data and the revenues I asked him to calculate in a separate column the market share of the players. And we got estimation in a separate column. So as you can see BlackRedWhite has a 9.4 market share. We also noticed that some of the players were missing, for example IKEA. So we asked him to add it and we got it as well. Then we removed the BlackRedWhite which we had on number 1 and number probably 8 and we got the following table. Then we remembered that actually there are other smaller players missing. We asked him to add PagetMeble and BydgoszkaMeble which is another player. Again he got some estimation of the revenues both in Polish currency and US dollars and we got the new table. The next thing we want to understand is to see what is the retail chain of those players. Some of them they have their own stores whereas others they don't. So I asked him to add the number of stores owned by the initial producer in a separate column. He got data for some of the players and we got them again in a separate column. The last thing we wanted to get is the revenue per one store to see how big those stores are. So we asked him to add a new column, revenues per one store, and thanks to that we were able to see which stores are bigger. So for example for IKEA we got that one store is generating roughly 100 million US dollars. And this is where we finished. As you can see pretty fast, it took us probably 3-4 minutes to generate this data. We get something which we can try to play with. Obviously the problem is that we tried it 3-4 times and we were getting a bit different results every time, both in terms of revenues and sometimes in terms of market share. Sometimes he got lost at the definition of the market share so he would use two different ways of generating them. But generally speaking as a starting point it's pretty good. The obvious problem is that you cannot always trust the data provided by the chat GPT. But generally speaking some of the things he got pretty accurate but you have to be careful with the data you're using.