Outside analysts, working for banks and other financial firms, using public/non-insider data, will write up a report about what they think the numbers coming out of Netflix's quarterly earnings report will be, to try to predict how the stock price will be affected by the report. This is why sometimes you see amazing, other-worldly quarterly performance, but the stock price doesn't move at all- because everyone had already anticipated the performance and that had already been priced into the stock price.
So if you have subscriber growth numbers - you look at where the analysts get something wrong as part of their calculation. Maybe they think there will only be 1M new subscribers, but in reality there are 10 million.
We suspected (more than a decade ago and the relevant bug has been long-ago fixed) that some research firm was placing orders at the start and middle of each quarter. These were suspicious orders as it was for a customizable product but the customer was not customizing anything (where the customization was the majority of the value). The orders were going to New York, always paid, never any chargebacks, no apparent credit card fraud, or complaints to customer care.
Looking into it, the only thing we could figure is that we were leaking an incrementing integer as part of our manufacturing process and the customer was apparently willing to buy small orders to get access to that ID (and thereby estimate order volumes).
We changed the process to not leak incrementing IDs and the orders stopped after a short time.
The leak was that we generated an auto-incrementing integer order ID and printed it on the inner pack label.
There was part of me that thought about fixing the situation by first adding an extra ~25% bump to the mid-quarter and ~35% to next start of quarter (by incrementing the ID column with a patch). We obviously didn’t, but it was fun to contemplate.
I actually admired the lateral thinking if it was a hedge fund doing research.
So if you have subscriber growth numbers - you look at where the analysts get something wrong as part of their calculation. Maybe they think there will only be 1M new subscribers, but in reality there are 10 million.