Quote:
|
Originally Posted by David - PG
Why on earth would a fluctuation in a time series variable describing a random event (signup) be ridiculous? It is absolutely normal for ratios to vary. The lower your average sample size (# of sales in a period) the higher the fluctuations.
The McDonald's around the corner might have 8000 customers on a given Friday night. The Friday the week after they only had 4000 customers. 50% drop. I guess they're being shaved.
|
yeah ratios vary and you have to take ratios based on roughly the same number of hits. you can't compare a ratio of 1:1000 or 1:100 if you only sent 1000 hits to get that 1:100 and 10,000 to get the 1:1000 but the mcdonalds analogy isn't really accurate from my point of view.
say traffic numbers are roughly the same. this means mcdonalds would get the same amount of customers, just half would not buy anything. i really don't think you can compare internet sales to brick and mortar sales but if traffic numbers are roughly the same, the sources are roughly the same its very frustrating.