In my company, we currently measure our progress and success via total revenue for the year. We sell a service that can be booked for the next year or current year. Thus, we measure a year's total revenue on a 104 week basis. So for example, 2014's total revenue includes all revenue earned from sales made in 2013 (weeks 1-52) and 2014 (weeks 53-104); the catch is just that the clients purchased the service to occur in 2014. This ALSO means that during 2014, sales are being made for both 2014's and 2015's total revenue. A file is attached with this data. Normally there would be numbers within the "2020F" forecast column, but I have deleted them out but kept the final 15 million end number that my boss has set as the end number for the forecast.
My boss has tasked me with trying to improve our current forecasting method. For now, the forecasting is just based on history, trends, and seasons and I will probably try to add more into it later. We currently take an average of the previous 3 years and then multiply by a growth rate to get our forecasted number for the weeks. To be honest, I am a little dumbfounded on what I could do to improve the forecast, so I am reaching out to those who might know better than I would. Using the forecast.ets function seems interesting, but I honestly cannot figure out how I could possibly format my data so that it will take other years into account properly.
I also have access to just the raw sales data with the exact date it occurred. I am currently trying to see if I can automatically have excel continuously count the weeks starting with our first sale and maybe that will work better, but in the meantime, I wanted to ask what you thought about this to see if there are any other good ideas out there on how to best forecast this kind of dat
Thanks!
My boss has tasked me with trying to improve our current forecasting method. For now, the forecasting is just based on history, trends, and seasons and I will probably try to add more into it later. We currently take an average of the previous 3 years and then multiply by a growth rate to get our forecasted number for the weeks. To be honest, I am a little dumbfounded on what I could do to improve the forecast, so I am reaching out to those who might know better than I would. Using the forecast.ets function seems interesting, but I honestly cannot figure out how I could possibly format my data so that it will take other years into account properly.
I also have access to just the raw sales data with the exact date it occurred. I am currently trying to see if I can automatically have excel continuously count the weeks starting with our first sale and maybe that will work better, but in the meantime, I wanted to ask what you thought about this to see if there are any other good ideas out there on how to best forecast this kind of dat
Thanks!
Attachments
Last edited: