# Pricing formula - Why power 2/3 ?

Lokad delivers forecasts and we charge by the forecasting task. Simply put, the more forecasting tasks you have and the more you pay, but with a quantity discount. This post is a small attempt to explain a bit the mysterious rationale behind the Lokad pricing formula.

### The power ^{2}⁄_{3} formula

Yet, if you have 10x more forecasting tasks, you do not pay 10x more with Lokad because we provide bulk “implicit” discount in our pricing formula. Indeed, the price formula is **(number of tasks) ^{2}⁄_{3}** .Although the power

^{2}⁄

_{3}might look complicated, the effect is quite simple:

**If number of tasks is x8 ==> total price is x4**

As other way to express this formula is to say (it’s completely equivalent)

**If number of tasks is x8 ==> price per-forecasting task is ^{1}⁄_{2}.**

Thus, if you are using 80 tasks, you are charged only 4 times more than if you were using only 10 tasks, although you are consuming 8x more tasks. If you are not comfortable by multiplying the number of tasks by 8, then you can also consider that this formula is equivalent to a 25% discount (roughly) if you take two times more forecasts. The discounts pile up. So if you multiply the number of tasks by 64 = 8x8, then the per-forecasting task price gets divided by 4.

### The reasons for the quantity discounts

Most software companies provide bulk discounts, but few of them would actually explain why. They are many different reasons for adopting such a pricing scheme. A common reason is that the marketing costs are lower if you sell 1000 licenses at once as opposed to sell 1000 licenses to 1000 customers. This reason somehow apply to Lokad but it’s only a small part of the explanation. The key factor when defining the price of a software is to figure out the benefit that the customer gets from using the software.

**How much benefits do our customers get from a marginal forecasting task?** Let’s consider the situation of a small retailer selling 1000 products. The first 50 products are top selling products, thus the retailer needs accurate sales forecasting for those products because those top selling products represent a really big percentage of the business. Then, the next 100 products involve moderate sales. It’s still interesting for the retailer to get forecasts but the benefits are lower because those products represent a smaller part of the business. Then comes the remaining 850 low profile products. The sales of those products are not really worth to be forecasted because those products represent only a small percentage of the business anyway.

From the Lokad viewpoint, it’s clear that Lokad cannot charge the same price for the sales forecasts of the 50 top selling products (that are really useful for the retailer) as opposed to the sales forecasts of the 850 low profile products (that are much less useful to the retailer due to the limited business impact). Thus, Lokad must charge less per-forecasting tasks as the number of forecasting tasks increase, because the per-forecasting task benefit of the customer is decreasing too.