October 19, 2016
Happy Holidays everyone! Well, that may be a bit premature, but not in the advertising world where the preparation for the largest sales days of the year begins early. With the holidays coming up, it is only right that we begin speaking to seasonality and how to bid differently based on seasonality.
First things first – let’s steer away from the holiday season for a little bit and just think through seasonal bidding in general. How do most users go about implementing bid changes? They take the historical data (last 7 days, last 14 days, last month) and make adjustments based on that historical performance. Makes sense, right?
Let’s dive into a model that works this way and see how it goes. Let’s say Carrie Albright owns an online store selling hideously overpriced hipster furniture. She really doesn’t see much difference in her sales week over week or month over month due to any sort of seasonality. So, week one her account sees a 500% ROAS due to $5 CPCs and gaining an average $25 revenue per click. The goal of her account is to max out revenue at a 300% ROAS – so she raises bids accordingly as Average Position and Impression Share show plenty of growth availability. From there she raises bids to $8.33 and she sees the 300% return with more traffic coming through because she got more aggressive. Makes perfect sense!
Now, let’s say across the web, Will Larcom sells tissues and hand sanitizer in bulk (Yes, the same Will Larcom who presented on Seasonality bidding in his recent webinar). He typically does monthly bid changes within his PPC account – so he notices from September 1 – September 30 he had a 150% ROAS as result of his $2 CPC and $3 of revenue per click typically. Will dreams of having a 500% ROAS as Carrie has – so he runs the math on these keywords that are all performing identically and decides he needs to set bids to $0.60 in order to achieve that ROAS based on September performance. What happens? Will’s account in October explodes with $20 per click in revenue because much more people are buying tissues and hand sanitizer when they click his ads comparably to September. He didn’t think about seasonality when making the changes he made, so he didn’t max out the traffic availability in the month of October. as a result he ended up with lower total revenue but a ROAS of 3333% with $0.60 clicks and $20 revenue per click on average. He looks back at last year and notices a very similar trend where he was receiving $1.50 per click in September 2015 and $20 per click in revenue in October 2015.
Now, PPC isn’t as easy as a junior high math problem as presented above – but the basis of these stories remains:
You need to take into consideration seasonality when bidding if your account shows that seasonality exists within your sales rates.
So, how do you go about doing this?
Below shows the situation Will was in with his bidding techniques. What is added here is a seasonal adjustment based on the previous year. In this case – the month over month differences last year from a revenue/click standpoint equated to a 667% differences. Meaning – if we got the exact same amount of traffic on each keyword – we can project that we’d receive 667% more revenue. Below shows this calculated into the changes and the new bid (seasonally adjusted):
This can be done week-to-week as well – just comparing the previous year and how revenue/click moved from week to week the year before. Assuring no major adjustments were made week-to-week in the previous year is crucial here – but as long as you put confidence in that, you’re looking at everything from a % standpoint. So, if September this year performed better than September last year – we’ll still be considering that in our bids for this year, but we’ll also be considering the differences in performance between the two months last year in order to assure we’re taking seasonality into consideration.
Now, back to the holidays. During holiday season the differences could be drastic in your revenue/click numbers on a much shorter basis. Black Friday to Cyber Monday could perform one way, while the days after Cyber Monday perform completely differently. So, let’s make a bidding schedule accordingly!
A few different things need to be considered when creating a holiday bidding schedule:
- Difference Between Revenue/Click Last Year in General vs. Revenue/Click During Holiday Times
- Day-by-Day Revenue per Click Numbers from Last Year
- Difference Between Day-by-Day Revenue per Click Numbers During Holiday Time Last Year and Pre Holiday Time Last Year
- Trends within those Differences
In order to see seasonality differences when the holidays hit you must visualize things as a whole. Then, you can assess things from a day-by-day perspective and find trends in order to keep a steady bidding schedule and have as few as days possible where you’re making drastic changes but ensuring it aligns with the results seen last season.
The chart below shows how this can be laid out in an Excel sheet:
As can be seen in this situation the holidays have strong points throughout the month of December – and points where performance isn’t as different as you might think. From here you can group the days as we did here and make decisions on when to adjust your bidding model and by how much you want to raise bids based on the seasonality seen in the previous year. Essentially, you use the performance from Pre Holidays in 2016 – and use a chart similar to the one above to adjust farther based on expected seasonality in late November and throughout December.
Overall using seasonality within your bidding strategies is extremely important – and, for the majority of eCommerce, taking that a step further into the holidays could help ensure that you maximize your revenue when your revenue per click fluctuates throughout the season.