Nov 282011

RKG Data: Q4 Strength Continues Through Black Friday

The pepper spray had hardly cleared the air before the first batch of Black Friday numbers started rolling in, offering signs of optimism for the larger holiday season and helping kick off a surge in the stock market.  ComScore reported online Black Friday sales were up 26% year over year, while IBM put the increase at 24.3%.  Figures for offline shopping, including those from the National Retail Federation and ShopperTrak, were also generally positive.

At RKG, we've been analyzing aggregate holiday paid search performance for a sample of clients that have been working with us for at least 19 months.  This is the same grouping of well-established programs that we've used for our quarterly reviews, which have generally tracked well with the results that the search engines report.  Here's our take on the holiday data so far:

Early Q4 Growth Continues Through Black Friday

For the first half of Q4 we saw a 43% YoY increase in non-brand paid search spend, and that growth has continued right through Thanksgiving and the Black Friday weekend:

Our clients' year over year increases in ad spend really seemed to ramp up towards the end of Q3 and revenue has kept pace.  On Black Friday, we saw a 43% YoY increase ad spend, which was right in line with our average of the first half of Q4.  Ad spend on Thanksgiving Day was up 42%, so we didn't see a dramatic change in shopping behavior within the week from year to year.

Interestingly, spend growth in the fourth quarter has been driven by click-through rate increases rather than higher costs-per-click:

On Black Friday, CTRs were up 29% YoY while CPCs were up just 2%.  While some paid search traffic growth is surely cannibalistic of offline sales, retailers are likely seeing PPC cannibalize organic search traffic to some extent as well due to changes the engines, particularly Google, have made to the SERP, which are drawing more attention to the paid listings.

Daily Trending

Looking at these results another way, with November 1st normalized to 100%, we can get a better sense of the spikes retailers see when holiday shopping really kicks in.  Again, here are paid search ad spend and revenue:

After several weeks with normal ups and downs, Black Friday ad spend jumps to 56% above 11/1 levels.  Ad revenue rises even farther, to 92% above 11/1 levels.  Taking those together, here's our view of return-on-ad-spend by day:

Generally speaking, as long as budgets are flexible and efficiency goals haven't changed, we would hope to see ROAS remain fairly constant even as traffic surges.  Here we see that ROAS did jump to about 23% above November 1st levels on Black Friday.  I should point out that these figures are not accounting for the time of click, so the ROAS jump may be the result of earlier clicks converting on Black Friday after promotions began.

Mobile Trending

IBM's Smarter Commerce report included a number of interesting data points around mobile traffic.  They estimated that 14.3% of Black Friday traffic was mobile, with the iPad alone accounting for 4.8% of traffic.  In our data we saw "mobile" traffic surge to account for 11.5% of Black Friday traffic:

I put mobile in quotation marks above because tablet users are generally lumped with smartphone users even though their shopping behavior, as we pointed out two months after the first iPad was released, suggests they are better grouped with desktop and laptop users.  Tablet time of day usage also suggests they are not commonly used on-the-go like a smartphone.  That said, we did see a significant spike in tablet traffic around Black Friday:

Since it accounts for about 94% of tablet traffic, above is our view of iPad's share of paid search click traffic.  We see the iPad's share jump from about 4% at the beginning of November to around 6% over Black Friday weekend.  It's possible consumers see the iPad as the ultimate shopping device, or maybe usage spiked due to so many of us traveling for Thanksgiving.  While normal tablet usage correlates closely with the hours most people are at home, we'd also expect tablet owners to take them along on trips.  With some predicting a prolonged surge in mobile+tablet usage in December, it should be telling how tablet share trends over the next week now that everyone's back to work.

Cyber Monday and Beyond

While this holiday season seems off to a great start, we'll have to remain cautiously optimistic about the potential for the next few weeks.  Last year also looked incredibly promising through Cyber Monday, but year over year growth fell back to earth shortly thereafter.  We've suggested before that holiday sales may become increasingly front-loaded as consumers become accustomed to getting the best deals on the big event days like Black Friday and Cyber Monday.

Here's a view into how that shift looked from 2009 to 2010:

Here ad spend is normalized such that three weeks before Thanksgiving (TG -3) equals 100%.  In 2010, spend peaked the week of Cyber Monday while 2009 spend levels didn't peak for another two weeks.

Here's hoping that this year looks more like 2009 than 2010.


2 Responses to "RKG Data: Q4 Strength Continues Through Black Friday"
Mark, great data & analysis. I don't think you have enough quarters of accurate forecasting under your belt to be a paid search investment guru just yet, but get another one or two quarters right and lots of suits will be buying you lunch... A couple points: 1) Your data's for paid search overall, but what interests investors is to know specific YoY growth numbers for Google. Can you share that data? I suspect your Google Q4 2011 data will show stronger YoY growth than for Bing/Y! because as you state, it's Google that's made big moves to increasingly direct organic eyeballs over to paid clicks. 2) I believe consensus estimates are for Google's revenues to grow 30.3% YoY in Q4, yet your data for the first half of Q4 shows 43%. Assuming your increasing front-loading thesis is correct, can you project for 2011 based on 2009-2010 Q4 trends to make a call as to what Google's Q4 YoY revenue growth is likely to be? [This will undoubtedly require looking at Google-only data, but the reality is you're already judging your quarterly data by its correlation to Google's reported revenues...] 3) Your front-loading thesis is an interesting one, and implies you'll be downwardly & significantly revising your Q4 #'s. Another Q4 force to contend with, though, is the fact that this Thanksgiving-to-Christmas season is one day longer than last year's (30 vs 29), and is in fact the longest T-to-C season since 2005. That should in theory add ~3% more to the YoY ppc holiday pie. Incidentally, during the last Great Depression, FDR moved Turkey Day up from 11/30 to 11/23 to stimulate the economy, detailed here on my blog: 3)
Mark Ballard says:
Hi Chris, thanks for weighing in. To start I should say that my goal with these types of analyses is not purely to predict the engines' performance, but I do make an effort to limit the data set to clients that should generally track well with overall industry trending (i.e. larger, established programs and those that have not significantly changed goals). Even though our numbers in these analyses have been close to the engines' the last two or three quarters, as any investment prospectus will say, "past performance is no guarantee of future results." To your points: 1) I don't have the engine splits at hand for the past week, but Q4 numbers through 11/15 did show Google making larger YoY gains than the overall numbers. I'm not sure if we'll put any of those out before the end of the quarter, but we will before earnings reports. 2) I won't wager a guess on Google's Q4 results now, but I would be surprised if they came out up 43% YoY or higher for the quarter. This could definitely be a quarter where our numbers, even using a consistent methodology, diverge significantly from Google's. 3) The front-loading effect is a tricky one to predict. Last year, it seemed to hit people pretty hard, so the comps should actually get a bit weaker moving forward. Obviously, in order for YoY numbers to take a hit this year, that front-loading must accelerate and it's by no means certain it will. I haven't looked specifically into the impact of differing T-to-C window lengths in past years, but I suspect the one day difference this year won't have a material impact on total spending for the season. Great post by the way, it sounds like FDR's move to extend the window didn't really work.

Leave A Comment