The following case study was first published on the SearchForce blog at https://www.searchforce.com/blog/bid-management-for-seasonality/.
We are all very aware that the four days between Black Friday and Cyber Monday is a critical sales period for many online retailers. Campaign performance during this time period can make or break a retailer’s success for the entire year. Applying the same bid management strategies during this period of escalated search volume and conversion rates won’t be able to exploit this seasonal trend to maximize sales. During this year’s holiday buying season, some retailers running on the SearchForce platform have leveraged dynamic bidding portfolios to react faster to the changing traffic patterns and maximize sales volume.
The Seasonality Challenge
Bidding algorithms are designed to consider historical keyword conversion rates to determine optimal bids to maximize sales volume at a specified ROAS. The data set they consider may look back over a few days to a few weeks of historical performance data to determine optimal bid levels for tomorrow. But the weeks leading up to Black Friday are not indicative of the volume, CPC rates, average order values, or conversion rates on Black Friday. Thus, we can count on the bidding algorithms to make incorrect predictions and drive sub-standard performance if left to “business as usual.”
SearchForce Client Solution
One innovative retailer employed an elegant solution to address the seasonality challenge by leveraging SearchForce’s dynamic bidding portfolios. Rather than letting a one-size-fits-all algorithm continue manage bids as during any other time of year, they designed a sophisticated portfolio set that automatically changed bidding strategies on specified days based on each keyword’s performance during the previous year’s holiday weekend. Sounds complicated. But surprisingly this solution only took a couple hours to setup and 10-15 minutes of maintenance throughout the weekend. Here’s what they did. . .
The advertiser maintained the existing base portfolio that was set to optimize to a 400% ROAS.
Black Friday ‘movers’ sub-portfolio
This portfolio was set to automatically extract any keyword from the base portfolio that received over 10 clicks and had a conversion rate of over 3% during the previous year’s holiday weekend. These became their “movers” keywords as they were all successful in driving converting traffic during last year’s Black Friday weekend. A bid rule was scheduled to run in the early morning of Black Friday which would automatically increase bids on these keywords by 25%. The goal of this portfolio was to rapidly adjust for the dramatic increase in competition between Thursday and Friday. This portfolio was deactivated after Friday bid adjustments.
Black Friday ‘selected’ sub-portfolio
A second Black Friday sub-portfolio was created to segment early-funnel head terms that they wanted to ensure a high impression share on. Keywords like “Christmas sweaters”, “leather gloves”, and “gift baskets” that attribution reports had shown to be critical early-funnel keywords assisting conversions on subsequent clicks. This keyword group was scheduled to have bids increase by 50% on Black Friday to ensure a top of page position and high impression share. This sub-portfolio was deactivated after Friday bid adjustments.
Weekend ‘below the fold’ sub-portfolio
The ‘movers’ keywords may not have experienced enough traffic on Friday to accurately assess conversion performance, but they would certainly have enough volume to identify a CPC/Position relationship. Therefore, low ranking keywords would be optimized by position during the remaining days through Cyber Monday. This sub-portfolio segmented all ‘non-selected’ keywords with an average position >5 on Friday only. Forget all other performance history, we’re only interested in Friday’s position at the current CPC bid. This keyword group was scheduled to have bids increased by another 15% since they were still showing below the fold. This sub-portfolio was scheduled to run daily through Cyber Monday.
Weekend ‘performers’ sub-portfolio
We expected a number of keywords would have received enough volume on Friday to optimize based on an ROAS target. Therefore this portfolio was created to segment any ‘non-selected’ keyword that received over 100 clicks and had a >500% ROAS on Friday. A bid strategy targeting a 400% ROAS was applied to these keywords with the response timescale set to short term and bid aggressiveness set to ‘aggressive’. Bids for these high volume keywords would once again be based on conversion performance. This sub-portfolio filter and bid strategy was left to run through Cyber Monday.
Weekend ‘selected’ sub-portfolio
Starting on Saturday, ‘selected’ keywords were switched to a position bidding strategy targeting position 2.5. As with the ‘performers’, this strategy was also set to a short-term response timescale and highly aggressive bidding. Knowing that an accurate analysis of subsequent assisted conversions would require much more time, the primary goal for these keywords continued to be based on position and impression share. This sub-portfolio was left to run through Cyber Monday.
All the rest
Note that any keyword not belonging to any of these dynamic sub-portfolio segments was continuing to be managed by the base bid strategy to a 400% ROAS based on standard historical data selections.
Post Cyber Monday
After automated bidding processes ran on Monday, all sub-portfolios were deactivated and base portfolio bid strategies were simply set to a short-term response timescale with the same 400% ROAS goal. This served to weight recent holiday keyword performance more heavily than performance from weeks before.
What were the results of this setup? Well, it would be anti-climactic to say that the retailer succeeded in driving more sales during the Black Friday weekend. We already knew that would happen. What’s truly exciting about this setup is the automation itself. Compare this automated bidding process to a 1st generation bid management platform. The most a 1st Gen bidding portfolio can do is segment keywords based on campaign structure and assign a single bid strategy that will be applied in all situations. But such a ‘one size fits all’ bidding strategy wouldn’t be able to identify and respond to the rapid volume and performance changes we knew would take place.
This 2nd generation bidding automation was both flexible and responsive to the dramatic performance changes that began on Friday. At around 3am each morning, keywords were being regrouped into dynamic sub-portfolios based on their performance during multiple time periods. . . initially based on previous year’s performance, and then on the previous day that ended three hours before. Depending on the performance grouping the keyword ended up in, the most appropriate bid strategies and bid rules were assigned and applied at the keyword level. Some keywords that remained below the fold on Saturday had bids aggressively increased a second time while others already hitting position thresholds were optimized to an ROAS target. And none of the campaign managers had to login to the account each morning of their holiday weekend to analyze performance and manually group and apply appropriate bid adjustments to ensure they were maximizing returns during the four-day period. No worries about a mistake being made, an adjustment being forgotten, or an analyst waking up too late.
Any large online retailer that depends on exploiting holiday shopping traffic can appreciate the benefits of such automation.