Let's say you've been logging all this data over time:
- everything you'd see in an apache log: IP, UA, referrer, date/time, req method, etc...
- each page view tagged to a session cookie.. to easily recreate user footprints
- log of all search terms used in internal (onsite) searches
- if using PPC logging ad group, ad used, bid, kw purchased, actual kw searched, referring site
- long term cookies for tracking repeat visits
- traffic from RSS feeds (which can be polluted by rss scrapers, btw. That could be good or bad depending on what you want to know.)
- a/b tracking data - ads, page copy, placement of elements, etc
- tracking of individual links on the page.. knowing which link/button on a page was clicked when more than option exists
- offline sources via unique URLs
- whatever else is possible.. ideas?
If all that is logged and you could generate any report you'd like, what questions would you like answered?
- all the dull and boring things typical stats programs already give you ;)
- conversions by key word searched in the SEs (organic and PPC)
- conversions by source (referrer and/or ad)
- referrals by key word searched in the SEs (organic and PPC)
- profit by kw
- profit by source
- ratio of spend to profit
- sales by time of day, day of week, etc
- top kwds and referrals by month for last 12 months (seasonal stuff)
- what pages of your site were crawled by who and when? which weren't?
- rankings (actually pg1 vs pg2, etc) of kwds used to find you in the SEs
- low ranking kwds that are generating conversions
- internal search terms, sorted by frequency
- last page viewed before checkout (the converting page - where they clicked the buy now button)
- most and least popular products purchased.
And that's just the tip of the iceberg, isn't it. Looking for reporting ideas here... what are some of the key bits of info to know that the typical stats software won't give you?