• TSG

    How long is the cookie expiration window for this client?

  • http://blog.efrontier.com sidshah

    I looked at transactions covering 2 months. Typically we see about 95% of all transactions in a 30 day window. So 60 days is a good length.

  • http://www.rimmkaufman.com George Michie

    Good stuff, Sid, thanks for the hat tip! There is some confusion in the literature about what is meant by the term “brand”. You’re using it here to denote that the user searched for a specific brand of apparel it seems. Often when people talk about the generic to brand search path they’re talking about the advertiser’s brand name, not the manufacturer’s brand name. In that context “Ralph Lauren” is only a brand search for Ralph Lauren, for every other retailer that sells Ralph Lauren clothing it is a competitive non-brand search phrase. Just wanted to make sure everyone was on the same page here.

    Also, your findings line up with ours in another sense as well. We find that customers who buy from more than one of our clients are disproportionately likely to use the same type of search phrase whenever they shop. Would be interested in your comments on this notion of user personas: http://www.rimmkaufman.com/rkgblog/2010/02/02/shopper-profiles-personas/ and http://www.rimmkaufman.com/rkgblog/2010/05/10/search-behavior-from-the-rkg-labs/

  • Henry Huang

    Hi Sid,

    This may be a silly question, but noticing your second graph (Starting and Ending Converting Keywords), you mentioned the high percentage along the diagonal. Seeing that 70% of orders converted after one keyword (from first graph), was the data for the second chart parsed of one touch-point conversions?

    Thank you,

    -Henry

  • http://blog.efrontier.com sidshah

    @Henry: In this example one click conversions are considered to have first and last click from the same keyword. I dont expect things to change much if we execlude one click conversions because the level of self assists (first and last keyword for a multi click conversion) in this example was about 80%. This is most likely due to Google Autocomplete.

    @ George: I like the idea of different personas but in search the lowest level of aggregation is the keyword/match type and the broad matched high volume keywords attract all kinds of users. So how does one action on it ? Any ideas ?

  • Henry Huang

    @SidShah – wow, that’s great to know. It is interesting to see that even in multi click conversions that 80% still have the same first and last keyword conversions. I would’ve thought that it would be less than that (maybe around 60-80% range). Thanks for sharing, I definitely enjoyed reading your post.