Launching usage-based pricing
Instrument, shadow-bill, then charge. The order matters more than the model you pick.
Most usage-pricing launches fail on data rather than on price. The meter is wrong, or it is right but nobody knew what customers would actually be billed until the first invoice went out. Instrument first and the pricing question becomes answerable.
Instrument first
Emit events before you charge for them. Metering costs nothing per event here, so there is no reason to instrument narrowly, and a month of real event data is what turns a pricing debate into a calculation.
Shadow-bill for a month
- Rate the events against the proposed price without invoicing.
- Look at the distribution, not the mean: the top decile is who will complain.
- Check the customers whose bill would move most, and decide before they do.
Then charge
Announce with each customer's own shadow figure rather than the price sheet. A customer who can see what last month would have cost them is being asked to check arithmetic, not to trust a projection.
Keep reading
Closing the month on the first
Why a close takes 9 days, and which of them are just arithmetic waiting to happen.
Anubhav DubeyArticleMetering that reconciles
Usage billing fails between the event stream and the invoice. Counting each line against its usage records closes the seam.
Anubhav DubeyArticleThe close is a pipeline problem, not a people problem
Nine-day closes are not caused by slow accountants. They are caused by 5 tools that each hold a different version of the truth.
Anubhav DubeyArticle