Quote:
Originally Posted by moviebiz
... month to month, year to year  lol
Testing timeframe depends of #of sales/day and #of billing's in cascade (IMO)
Sites doing 1 sale/day(1 billing in cascade) and sites doing 100 sales/day(3-4 billings in cascade) must be tested in different timeframes. |
wrong. although there's not a single formula, but many different formulas depending on variables, the whole idea of split A/B testing in online business is to work with short time spans, usually no longer than 2 months or year quarters (ie seasonal variations, holidays, etc). Online business has an extremely high development ratio and what was good 1 year ago may not be the same today. Think about the tubes scenario, would you say business is the same today than 1 year ago? let alone 2 years ago or more.
Someone could say "well, look at The Hun with the same design for years" and it's right, once you find something that works, you'll try to stick to it. Nevertheless, The Hun lost almost 50% of its traffic from 1 year ago to day.
about variables, the "trailer/no trailer" is just one of many. You can use color, style, navigation, layout, amount of content, type of content, shapes, sizes, whatever. All of them will affect the outcome in one or another way, some will have a bigger impact than others (color, text and content type are usually the most important ones) but all of them will have some sort of impact, and that's why A/B testing is needed (I prefer A/B/C to the very least, with C being a combination of A and B)
and finally, a little secret: EYES. any good designer or marketing person will know the power of eyes in a picture and also adapt it to the design needs. Most of the times designers should change them with something as little as 1px variation, but the results are always outstanding