Becoming a Forecaster
How well am I doing? How well will I do?
I have a background in marketing and business intelligence; I know what I want from sales data on books. What’s worse is, it’s available, just a bit too expensive for me.
What am I trying to do? Firstly, I’m trying to tell from the sales of my book so far, how well it will do eventually. Secondly, how sales will change when the second book in the series is published. Thirdly, how well will the second book do.
This sounds a bit like gazing at a crystal ball, but it’s a science called forecasting. And once you become a forecaster, you often start to refer to it as gazing into the crystal ball. But it’s fun, and never more so than when the subject is close to your heart!
So what do I know that will help me in this task? I know my Amazon and Kobo sales numbers, which I read every day and put into a spreadsheet. And… that’s it for hard data.
I used to be able to look at TitleZ, which gave sales figures for printed books and early ebook sales, exclusively from Amazon. That would be useful, because Amazon handles about 70% of all book sales. It’s less useful now because Amazon stopped allowing TitleZ access to ebook sales, and last time I checked, TitleZ was having difficulty reading anything from Amazon apparently. It’s of limited use anyway, because ebook sales now swamp printed sales, especially in the niche I’m in.
I can look at Amazon itself, which provides rankings of books. However, they adjust these rankings with weighing factors, including how long a book has been available and how many previous sales it has, and they don’t release sales to anyone with a browser. Still, I can track my book sales vs rankings and get a general idea, for example, of how many sales are being made by a title that’s in the ranking I was in last week – i.e. I can say that that title is making approximately the sales I made last week. But for this to be useful, I’d need to have tracked other books over time, and I haven’t.
However, there is a service which tracks book sales. It’s by Nielsen and it’s called BookScan. That’s what I want and can’t afford.
What would I do with it? I would use a method called ‘Launch Analogue’ to predict how well my book will do on the data I have so far. I have two months sales data so far, and the basic concept of a launch analogue is to find the most compatible matches to my sales numbers in an earlier launched book in the same area. Then I can draw charts and say these are the highest and lowest expectation I have based on similar books.
That’s only moderately accurate. There are too many factors coming in to be able to forecast with certainty – internal factors such as advertising, publicity and reviews and external factors such as holiday seasons or JK Rowling inconsiderately launching a book that absorbs 2.5 million * $20 of disposable income on the first day!
What’s much more interesting from launch analogues is that it would show the effect that a second book in the series has on the sales curve of both books.
Given that I have none of that, what can I do with just the hard sales data on my own book?
It’s already in a spreadsheet, so I can forecast future sales on the basis of sales so far. This is based on the concept that if 20 people discovered the book and bought it yesterday, another 20 may do so today, and if 21 people buy it today, then I might predict 22 sales tomorrow. I take averages – the average increase in sales over the last week and apply it forward. None of that is contentious, and the boundaries, i.e. the limits on the number of people who buy books of this sort are a long way off. I’m not ignoring factors such as good or bad reviews, but for an indie author, this is more external than internal – I can’t make people review it and I can’t make them give me good reviews. I could purchase reviews or sock puppet reviews but I will not. But that’s for another post!
Excel is the most widely used spreadsheet by some humongous margin. It’s probably the most widely used basic tool for forecasting because it has a function designed for the job.
This is called (predictably enough) FORECAST. You tell it the series of numbers you would like to base your forecast on and it does. It uses the least square method (look it up). It’s a widely accepted mathematical method of forecasting. And I LOVE it, because on my sales data so far, it predicts sales of my book will have absorbed the entire GNP of the western world by the end of the first year. Yay!
So, reality intrudes and I restrict the forecasts to what I believe are reasonable, and limit myself to the first year of sales.
My milestones are: 10,000 20,000 100,000.
10,000 is what I believe from my own predictions on my own sales data. That’s a pretty damn good result for an indie author’s first book. I’m basing that on sales continuing for the remainder of the year at the rate they have reached after two months. If I don’t make this, I won’t feel I have failed. If I do make it, I won’t feel I have really succeeded. But I will be happy.
20,000 is the number that was casually given to me by an agent as defining a worthwhile book in publisher’s terms. My forecasts reach this if I take the month on month increment as continuing – i.e. I sold 100 books month 1, 500 books month 2, then month 3 will be 900 books. If I make this, I will feel I have succeeded and I will be very happy. This is an optimistic forecast, because month 1 is always going to be low, so the step from month 1 to month 2 is always going to be high.
100,000 will get me on the NYT best selling list. You can imagine my reaction to that!
I have no idea how interesting this is to y’all out there. If you want actual data and mathematical formulae, tell me using the contact form and I’ll do a more detailed update. In any event, I will do a shorter summary of the position at significant points – release of Hidden Trump (book 2 in the series), end of the third, sixth, ninth and twelfth month of sales for Sleight of Hand.