Dec 32008

Online Retail Stats: CyberMonday 2008 vs. 2007

One day doth not a retail season make, but I thought folks might find it interesting to see our clients' "CyberMonday" results this year compared to last year.

"CyberMonday" is the first Monday after Thanksgiving. This year, CyberMonday was two days ago, December 1st, 2008. Last year, it fell on November 26th. The day was named by and is a large (but often not the largest) online retail sales day of the year.

Background: we serve about 100 online retailers, predominantly B2C, with an aggregate annual search spend of $100 million.

Considering only those clients we served in both 2007 and 2008 (to remove the impact of our firm's growth), and then aggregating all those clients together, and comparing 2008-12-1 versus 2007-11-26, here's what we saw:


Let's index those against the change in adspend:


We see sales grew much slower than adspend: our clients typically opted for lower efficiency targets this year versus last. Clicks grew faster than adspend, indicating lower CPCs, reflecting lower SPCs. Orders and clicks grew at the same rate, indicating this year's soft sales were more due to lower order size than due to reduced conversion. Impressions grew faster than clicks, indicating the engines earned lower effective CPM rates on their inventory.

We're privileged to manage search for some of the web's largest advertisers (many thanks for your business!) When we aggregate cost and sales across our clients, our largest clients, by definition, have the largest impact on the totals.

To avoid this skew, we also computed year-on-year CyberMonday metrics for each client (again, only considering clients we served in both years).

The following chart and histograms are based on this "equal-client-weighting" perspective. This treats each client equally, so a small client up x% counts the same as a huge client up x%. The "equal-client-weighting" provides a better sense of what a typical client experienced than provided by the agency-total metrics.


The means are quite different from the medians, a sure sign of skewed distributions. Here are the histograms themselves. The x-axis is % change, year-on-year.

Sales: Our typical client (treating all clients equally) saw less revenue this CyberMonday than last, but many also experienced healthy revenue gains. The story is quite mixed.

Advertising spend: Again, a mixed bag. About half of our clients reduced ad spend in response to lower sales-per-click arising from lower average order values. And about half our clients spent more on advertising, in some cases considerably more.

AOV: Nearly all clients experienced a considerable decline in average order. This trend began in early November. Very few clients saw increase in AOV, year-on-year, for CyberMonday.

Clicks: While some clients bought fewer clicks, most clients bought significantly more this year.

CPC: Nearly all of our clients instruct us to bid to a target efficiency, investing some predetermined percentage of sales (or margin, etc) in advertising. This approach mandates bidding lower when clicks perform less well. With many of our clients experiencing AOV softness, many of our clients brought down bids accordingly.

CPM: While paid search clicks are sold on a cost-per-click basis, CPM (cost-per-thousand-impressions) is a key metric for the search engines, because CPM indicates how well a job they're doing monetizing their traffic. On CyberMonday, most of our clients paid less per impression this year compared to last year.

Lots of numbers and charts. So, what does all this mean?

If our client base is a representative sample of online retail as a whole, we'd say that:

  • For most retailers, CyberMonday 2008 was not as strong as CyberMonday 2007. However, results were mixed. A decent fraction of retailers experienced strong year-on-year growth.
  • On average, retailers traded bottom line for top line, increasing their advertising aggressiveness.
  • The search engines experienced lower CPMs, pulling less revenue from their advertisers per PPC impression.

Honestly, before running these numbers, I expected to see far far far worse results.

While the online retail 2008 holiday season isn't starting out strong, it also isn't starting out as a total disaster either.

While we shouldn't extrapolate too much from a single day, I'm viewing CyberMonday 2008 as promising start, a tidbit of positive news about online retail and perhaps in turn about our national economy. May this season be a strong one for us all.

What are you seeing?

Update: ComScore put out a press release reporting overall strong CyberMonday sales: E-Commerce Spending Jumps 15 Percent on Cyber Monday to $846 Million, the Second Heaviest Online Spending Day on Record


12 Responses to "Online Retail Stats: CyberMonday 2008 vs. 2007"
Stephen says:
Alan, this post is yet another great example of why your blog should be "must reading" for anyone who considers themselves a professional search marketer. I forwarded this highly instructive analysis to my staff. Thank you for providing an example of how to "carefully" do year-over-year comparisons with substantive, easy to understand data-driven commentary. As an added benefit, it also provided excellent insights into the inter-workings of the "gears" of SEM. Thank you for sharing!
David says:
Good post Alan. This is one of the places I look for 'bigger picture' trends these days. You should do a follow up to the recession post you did last year. As it turns out, I think your blog came out right around this one's official start :)
Chris says:
The other thing to keep in mind is that this holiday shopping season is 15% shorter than last year's due to the fact that Thanksgiving was Nov 27th this year and Nov 22nd last year (27 days between Turkey Day and XMas this year, vs. 32 days last year). So on top of the fact that your CyberMonday data is showing weak results for the search engines relative to expectations, they'll have to contend with 15% few shopping days this year. Great post, Alan.
Glenn says:
Great post Alan. What kind of an affect do you think the 2008 calendar played in all this? The fact that CyberMonday also happens to be the first Monday in December. There is a shortened shopping season this year. Just curious, has RKG seen a sales drop for clients the Tuesday after CyberMonday vs. prior year? Just wondering if we--and the media--conditioned everyone to go bargain hunting on CyberMonday then they retreated back to 'recession shock' the following day?
Thanks for the kind words, Stephen & David. Chris, excellent point about the calendar. Spot on. To Glenn's point, I don't think the calendar (that is, fewer days this year between Thanksgiving and Christmas vs. last year) played any role in Monday's sales. No data to prove it, but I don't think most people plan ahead all that much. The shortened season will likely reduce total holiday sales -- that is what I recall happening when I was in cataloging. Glenn, haven't looked at "CyberTuesday" yet (yesterday) but will this week and drop you an email. Of course, extrapolating from a single day isn't terribly solid. But, since CM gets so much press, and since the media and the markets are hanging on every scrap of news about consumer sentiment, I thought it interesting to share what our agency saw. I've seen some items go by in the blogosphere in the last day -- the survey from NRF indicating sales not as weak as people expected; the ComScore numbers -- and also heard those reports being dismissed as biased. As if anything that isn't dire must be false... Online retail sales are weak, yes, but they have not imploded.
Glenn, to your point, several of our analysts have pointed out that in many ways our clients are influencing shopping behavior with exploding offers that end Cyber Monday, so yes, I think there is some of that in the numbers. Ultimately, like any benchmark, these numbers are fascinating, but not really actionable. All that really matters to any retailer is what's happening to her/his numbers and how should they react. George
Matt says:
We saw almost double sales this Monday vs. 2007. Of course, we did have a site wide promotion that we had been running 8 days prior which ended on Monday. Tuesday and Wednesday have both been lower than 2007. Looking at all 3 days combined we are essentially flat.
Don says:
Just discovered your blog and loved the two posts I have read. We are a small multichannel retailer (unfortunately too small to be a prospect for RKG, :))and are in the small group that had great results over the holiday weekend. Our customer base is primarily B2C. Our CyberMonday online sales were up 44% YOY on a 3% increase in traffic. The week prior, which ended this past Sunday, we were up well over 100% vs last year on a meager 1% increase in traffic. This year, we were very aggressive with our offer for the holiday weekend and it paid off. We also sent our Black Friday email a day earlier than last year and then sent a reminder on Sunday morning. Sunday was our highest traffic day of the year so far. The question that remains unanswered is did we rob from Peter to pay Paul.
Matt, Don, thanks for your feedback. It is a complicated puzzle! The exploding offers both cannibalize later sales and reduce the net margin on each order, but they also create urgency and thereby generate incremental orders, too. But how much of each of those factors? Hope the rest of the holiday goes well for you both! George
Jeremy says:
Those offers may also serve to bring in new customers who have a positive experience and will continue to shop with you in the future and recommend you to friends. At times you may want to look at discounting and special offers as a marketing expense that can be justified if they bring in enough new blood. Try to build lifetime value into your acquisition strategy; a new customer is worth more to you than just the first sale.


Check out what others are saying...
[...] as we’ve pointed out before here and here, despite the sky-is-falling stories in the popular press, we’re not seeing utter collapse in [...]
[...] More about RKG here, more holiday 2008 PPC data here and here. [...]

Leave A Comment