How much data do we need to confirm that the new site design isn`t going to convert well?

Selected answers from the Dumb SEO Questions Facebook & G+ community.

• 
  Shawn Cohen: A client launched a new design recently and conversion rate has tanked. I wasn't a fan of the new design but that's another story.
  
  Question: how much data do we need to confirm that the new site design isn't going to convert well and that we should start A/B testing various layouts? BTW, they won't A/B test unless we have data showing that it's absolutely necessary. Thanks for your help.
• 
  Jeremy Estes: This happened to us when we launched the new waspbarcode.com. certain leads dropped way off.
  
  after the first month, I knew we had to make changes to the paths.
  
  rather than introduce a radical layout change, try testing on something like action buttons or headline changes. Make sure to keep original elements in your testing.
• 
  Wakefly, Inc.: +Shawn Cohen ;how long has the new site been collecting data?
• 
  Cristian Oana: If you have data from before than the minimal sample size should be one week. ;
  
  Then you have to make the data sing your tune. If the site redesing is launched during your busyest seson then the sales no might show higher levels then previous weeks so you'll have to compare the week year over year.
• 
  Shawn Cohen: Yeah, we definitely have YoY data for the past several years. We are in a busy season at the moment and that is making up for the lower conversion rate in sheer revenue but waiting a month and tracking it daily until then sounds like a good plan. Thanks everybody.
• 
  Greg Baka: +Shawn Cohen ; if you are doing true A/B testing you probably want to cycle through at least 200 visits over at least one full week. A thousand visits over a full month would be even better mathematically.
• 
  Shawn Cohen: +Greg Baka, yes that sounds like a good plan. We get a couple thousand visits a day so the volume won't be an issue, thankfully.