Naive subtraction model (9/11)
Well, this wouldn't be so interesting if we didn't touch the estimation and calculation part here. One might wonder - can we estimate how much of an effect it had? And the answer to that is - yes! Kinda... So the simplest way to do that is to take those "spikes" we found, calculate them precisely, and subtract them from the review dynamics to get a new estimate. In the chart you see two curves: the higher one - the actual amount of reviews over time, and the lower one - the subtracted variant. That alone gives us a 13% of change and a total of 4300 reviews acquired. We can call this a "guaranteed surplus", and as you can easily see - it is linear, the "gap" between them is proportional. But we know, since Steam uses some feedback algorithm, this model is clearly under-counting.
0/0