If you manage paid search marketing and your business is retail, the weeks between Thanksgiving and Christmas probably don’t give you much time for sober reflection. Retail is intensely seasonal, and that is as true online as it is in stores. Search engine marketers generally understand the pattern, but they vary widely in how they choose to respond to it.
One response is simply to ride the season out – to assume that competition among retailers will intensify during the holidays, driving everyone’s paid search bids higher, but that otherwise competitive patterns won’t deviate much from non-holiday conditions. This approach assumes that during the holidays, and other key buying seasons, the crucial difference is one of volume. Defending an established search market share will get more expensive, but other changes in the market dynamic will even out over the course of the year.
But such a strategy overlooks the increasingly frequent pattern of market disruption. A new entrant in any given segment can dramatically alter the landscape. Some retailers, especially new players attempting to elbow their way into the market, are likely to bid more aggressively during seasonal buying spikes, because so much is at stake during these periods. They are unlikely to signal their intentions in advance.
Qualitative changes in the market can come about quickly and unexpectedly, and this becomes more likely in the lead-up to important buying seasons. Right now is a good time to examine the systems you use to optimize your search bidding, to see whether they are responsive enough to meet your objectives, during the holiday season and beyond.
There are important differences between bid optimization systems. You should view yours not simply as a productivity tool, but as a competitive weapon. The tool should enable you to predict the actual value of every click, and to quickly place the right bid to secure that value — not just a daily average bid that will get results at certain intervals during the day, but the optimum bid for every hour of every day.
Seasonal market fluctuations are more complex than many online merchandisers realize. In the first place, there are multiple “holidays” and buying seasons disrupting the ordinary patterns of commerce.
Figure 1 lists key planning periods for retail in the last quarter of the year. Clearly, the season is more complex than a peak in shopping on the Thanksgiving weekend. Each of these events leading up to Christmas has its own pattern of searching and buying, and for products that involve longer decision cycles, a “conversion” attributed to a click may mean one thing in October and something else entirely on Cyber Monday.
And this is only for Q4. Key buying seasons occur year-round. Florists, for example, depend heavily on demand around Valentine’s Day and Mother’s Day. Accountants look forward to tax season — as do travel-related businesses, since many people spend their annual refunds on vacation splurges. There are many other obvious examples.
It isn’t enough to know when these seasonal spikes will occur. The impact will be complex. Experience may suggest to a retailer that parents begin hunting for back-to-school items by mid-July and plan AdWord campaigns accordingly… only to discover a lag in activity that makes intensive July bidding futile.
A high-frequency bidding model
Marketers focus a lot of attention on the ramp-up before the holiday and the day of the holiday, but it is important to account for what happens after the holiday as well. Often, SEM managers who rely on daily bidding are slow to adjust to the fact that the holiday is past. As buying activity settles back to normal patterns, the advertiser will want to pull back on the bidding, but do so efficiently.
From hour to hour, there are discernible patterns — times of day when click volumes tend to spike, or when higher percentages of clicks to lead to conversions. This is the case on ordinary days, as well as the days leading up to a major buying holiday, but the holiday pattern is almost certain to change, in ways that may be unpredictable.
Two years ago, analysts saw a completely unexpected spike in online activity the Sunday before Thanksgiving. Most retailers were caught off-guard by the amount of research and buying that took place, even before Black Friday. This surge, which some observers dubbed “Orange Sunday,” turned out to be a one-time event. But it demonstrated that predicted seasonal buying patterns can be volatile, punctuated by surprise spikes.
This past year, retailers saw a big change in the way devices were used, as consumers acquired experience and confidence in using smart phones and tablets for shopping. Thanksgiving, Black Friday, Super Weekend and Cyber Monday all saw extremely heavy volume on mobile devices compared to previous years.
These examples demonstrate the need for agility in SEM. The Orange Sunday spike and the rapid growth of search volume via smartphone each represented an unexpected opportunity for paid search marketers to seize — if they were equipped to detect and respond to these events in real time.
Real time increasingly means hour to hour, not day to day. When the next Orange Sunday occurs, SEM managers who use daily average bidding will be unlikely to notice it until it’s past. Those equipped for High-Frequency Bidding will be prepared to make time-of-day predictions of the value per click (VPC), around the clock, and will be ready to adjust their bids quickly when competitive activity or consumer demand abruptly changes.
In short, online retail traffic patterns change constantly, and that change is accentuated during holiday periods. High-frequency bidding makes the advertiser more responsive to changing patterns, not just from day to day but during these days of intense activity. Figure 2 illustrates how these patterns can change from hour to hour and at different points in the buying season.
Strategy and bidding
There are two ways to look at a marketing budget: You may be asking, “Am I squeezing every possible dollar out of this budget?” Or you may be focused on a non-financial objective, such as position — your goal is to have your ad in the first position throughout the period. This kind of objective is typical for an upstart player trying to establish a market foothold. Most established retailers, however, are focused on financial goals. They want to spend less and generate as much revenue as possible.
An established player will try to optimize bidding toward a goal such as:
- Generate more revenue from existing budgets.
- Reduce costs while at least maintaining current revenue.
- Maximize profit for the accounts/campaigns.
Goals like these typically are expressed in terms of Cost per Acquisition (CPA) or Return on Investment (ROI). The retailer can only control certain things: keywords, keyword expansions, the ad copy, personalization of the experience on landing pages, device bid modifiers (allocating bids across various customer device platforms), and, of course, the bids. The retailer does not control competitive activity or patterns of customer activity on different devices.
Nor does the advertiser have control over the actions of the search platforms. During the holidays, the platform vendors test out new ways to deliver data. For example, during past holiday periods, Google would remove standard plain-text search ads and present primarily Shopping ads, which might take up the entire first page. Many of these experiments have been unannounced, and they challenge advertisers to respond quickly, so that they can get the most from their bids.
The complacent bidder may assume one daily bid will cover the highs and lows of value per click effectively, averaging between the value achieved during high- and low-demand periods to achieve an adequate return on the day’s bids. Figure 2, above, suggests otherwise.
The two horizontal axes show the day of the week and time of day; the vertical axis shows the number of clicks for a specific keyword. Visualizing activity in this way, and knowing that the technology exists to analyze auction activity and place optimized bids several times an hour, it should be clear that average daily bidding leaves a lot of money on the table.
In addition to clicks, the analytics for high-frequency bidding can project the cost per click (CPC) from hour to hour. Other analytics can estimate the likelihood that clicks from a given time of day will result in conversions — a critical measure of value per click.
A formula for bidding value
An effective strategy and technology platform for high-frequency, time-of-day bidding will incorporate all of the following:
- Time-of-day predictions. The platform can show you how the value of clicks varies throughout the day.
- Auction predictions. The provider can help you understand auction prices in order to optimize your keyword bids (and device bid modifiers) based on time of day/day of week.
- High-frequency bidding. The system allows you to act on those predictions, bidding multiple times per hour, as opposed to conventional solutions which only change bids daily.
Adopting a high-frequency bidding strategy is not difficult, but it is a change of process. Process change often is challenging. For this reason, it makes sense to begin the adoption process with a head-to-head A/B test, to demonstrate the incremental value of high-frequency bidding.
The holidays provide an occasion to evaluate how you’re bidding now and consider whether high-frequency time-of-day bidding would be worth investigating as a way to wring more value from your paid search investment through all the ups and downs of the season.
To learn more about Predictive Search Bidding or 7’s Customer Acquisition Cloud solutions, click here.