For some advertisers not all hours of the day or days of the week are created equal. Some companies see very apparent and specific times where search engine marketing (SEM) performance changes throughout a day, week, or month. With new sophistication in reporting capabilities, advertisers are now able to go beyond seeing only when traffic to a website spikes or dips. Through some technology platforms, it’s now possible to see times of the day, week, or month where conversion performance changes. This information sets the stage for advertisers to become more intelligent in making decisions on when to spend budgets in the most cost effective manner.
Instead of spending equal amounts of budget throughout a given time period, advertisers can determine, for example, that Saturday hours 12-15 convert at a higher rate than hours 1-5. From there one can bid differentiate between those hours in a day by increasing the base bid by a certain percentage when conversion spikes and decreasing when it drops. Some leading technology platforms even allow their clients to dynamically adjust bids based on DOW or TOD data. This will ensure that more budget is spent when it yields the greatest return and vice versa. This type of data can also be valuable in establishing budgets and setting bids for peaks and valleys in an advertiser’s seasonality. In looking back at historical daily, weekly, and monthly trends, it’s possible to verify, for example, that conversion rate spikes on November 20th and then drops back to normal levels again on January 7th. Utilizing this type of information empowers advertisers to make more intelligent intra day, week, and month bid and budget decisions. Many refer to this type of strategy as day parting or ad scheduling.
In order to create a day parting strategy one first needs to have access to at least four to six months of time of day (TOD) and day of week (DOW) performance data. If one is using this to create a seasonality strategy it’s usually best to use at least two years of historical TOD and DOW data for that time period. For example if a retailer wants to know when to increase bids for next year’s fourth quarter high season then they could use data from October – December 2008 – 2010. It’s necessary to use a large data set to ensure that the strategy is being built off of statistically significant trends. It’s also essential to exclude any time periods where your account produced unusual results as this will throw off the trend line. Examples of this could include a site relaunch or any sort of testing. Trademark terms are usually excluded from bid differentiation as they are generally managed to ad rank 1 at all times. Once the data set is in place one is able to compare the conversion rate for each time period (TOD, DOW) to the average for the overall data set. Once ratios to the average conversion rate for each time block are calculated, it should become clear if there are spikes and dips in performance over specific hours, days, or weeks. There are a number of methods to formulate the percentage to which a bid will be increased or decreased but many advertisers use the ratio to average. For example if the overall average conversion rate is 4.25% for the data set and on Mondays from hours 4 -7 conversion rate is 6.00% then the bid would be increased by 41% (6.00%/4.25%=141%, 100% = zero change). Some platforms have this step automated to the point where the tool recommends specific adjustments and pushes them out. Below are two examples of fictitious data sets where bid differentiation may be appropriate.
Example 1: Bid Differentiation for Seasonality Trends
Date Range: 6/1 – 8/31, 2008 - 2010
This fictitious example illustrates an advertiser that has historically yielded higher than average conversion rates through the end of August but has pulled back on budget in early August. This type of information can assist an advertiser in planning for next year to ensure that there are no neglected conversion opportunities.
Example 2: Bid Differentiation for Daily Hourly Trends
Date Range: Past 6 months of daily/hourly performance
This second fictitious example illustrates TOD and DOW data. It’s clear that there are spikes in conversion rate late on Mondays and Wednesdays, mid day on Thursdays, and all day Saturday’s conversion rates sit slightly above average. This could guide an advertiser to change the way they budget by increasing bids during these times to capitalize on increased performance. To offset total costs, bids could be decreased during times where conversion rate is below average. In theory overall spend would generally stay the same but because more of the budget is spent during times of greater efficiency more conversions would be generated at a more resourceful overall cost.
In Sum, bid differentiation can work best for clients who have explicit performance changes throughout different times of the day, week, month, or year. The degree of the conversion rate change over a given time period will affect the impact this strategy will have on an account. It may take some testing and tweaking of bid percentage changes over time to meet a perfect balance. Reporting closely on results and running in a true test and control environment will be imperative in truly determining how bid differentiation influences performance.
