First on an online retailing forum I follow, and later on the Shop.org blog, some online advertisers voiced concern late last week over their holiday 2006 pay-per-click campaigns: conversion down, costs up, traffic down. Other retailers chimed in to disagree — in their perspective, the PPC channel performed well: they enjoyed strong conversion and/or traffic and/or sales during Holiday ‘06.

Our data support the bullish perspective.

Across our clients, most experienced extremely strong Holiday results. On average, our clients enjoyed a 41% bump in sales per click after Thanksgiving. The largest holiday sales per click bump among our clients was an astounding 630%. Very happy holidays, indeed!

While 84% of our clients enjoyed SPC increases, 16% of our clients did not, experiencing flat or declining SPC rates. Digging into these cases, some were B2B clients selling business products with no year-end lift; others were B2C clients selling summer-season gear; and some were clients who opted to take a more aggressive position with their advertising during this period, generating more total revenue but albeit at intentionally lower SPCs.

Looking back at similar sales-per-click holiday bump rates from ‘05 supports that 2006 was strong holiday for pay-per-click driven sales.

So, stating the obvious, our experience suggests the pay-per-click advertising channel performed well in terms of conversion for most advertisers during Holiday 2006.

Update: Another article on Holiday 2006 click quality here

Methodology: we used sales per click as a metric, tracking how this measure changed from the 30 days prior to Thanksgiving ‘06 to the 30 days after Thanksgiving ‘06. We included both paid search and comparison shopping engine feeds.

For example, suppose that in the 30 days prior to Thanksgiving, retailer “X” bought 500K clicks which generated $1 million in tracked resulting sales; and in the 30 days after, retailer “X” bought 750K clicks and generated $2.25 million in resulting sales. Sales rose from $2.00/click to $3.00/click, yielding a holiday SPC click of 3.00 / 2.00 - 1 = 50%. If this example was actual data, retailer “X” would be counted in the green “45% to 50%” bar.

A few notes on this analysis. First, we count each client equally in the histogram — we did not weight the data by clicks. We did this because our interest was in the fraction of advertisers who might be experienced lower SPCs during Holiday, regardless of size. Second, our choice of metric (sales per click) by definition considers only revenue, not cost or profit. Rest assured: high SPC bumps drove similarly happy bumps in profit. Third, these results pertain to our client base, and may or may not generalize to the entire web.