Facing the prospect of being snowed in this Valentine’s weekend? Don’t worry, you’re not alone. With The Weather Channel predicting an intense snowstorm carrying “the coldest air this winter” for the North-East this weekend, advertisers should grab this niche opportunity with both glove-clad hands.

Take your optimization to the next level by not only targeting people in snowstorm-affected areas, but by further tailoring the Ads you show to these users by the current weather conditions, temperature, or time of day. AdParlor’s Weather Ad Scheduling, which is powered by our Automation Assistant, allows for painless automation of such logic. It works by automatically triggering actions based on criteria that are met. For example, you could configure rules such as:

  • “Is it currently snowing in Boston, MA?”
  • If yes, then activate your snowstorm-specific Ads and pause your standard Ads
  • If no, then pause your snowstorm-specific Ads and activate your standard Ads

The use of weather-tailored Ads vs generic Ads to further drive results has been proven effective in a case study led by AdParlor for Molson Coors with great success, with up to 89% improvements in key performance indicators such as Link Clicks.

Automating your optimization based on the weather conditions is just one such example on how you can be more efficient and effective with your advertising. We’d like to share with you a piece from our “Best Practices & Optimization” support library that explains the full flexibility of the AdParlor Automation Assistant, and how you can use it to drive better results with less effort, snowstorm or sunshine, all year long.

What is optimization?

“Optimization” has become a meaningless buzzword in the advertising space, tossed around by advertisers and agencies alike – but what does it actually mean?

From a mathematical perspective, optimization can refer to finding the combination of variables that lead to the ideal point (usually a maximum or minimum) of a function which describes something of interest. As a very simple example, let’s say you are the owner of an ice cream stall. Assuming all overheads are constant, your revenue (f(x)) can be modelled as a function of the individual price (x) you charge for each ice cream; or in other words, how much you make in revenue depends on your selling price. If you charge too much, no one will buy ice cream from you; but if you charge too little, then you miss out on potential revenue. Finding the selling price at this tipping point enables you to get the maximum return on your investment.

In the advertising world, this is much more complex. There are many, many variables which influence your metrics of interest (e.g CPA, total revenue, CPM, inventory supply/demand etc); these variables are also likely to depend on each other, and constantly fluctuate with the market conditions. Defining an accurate objective function to represent the whole space would not only be time consuming but extremely difficult; thus, optimization in the traditional sense is not of great use here.

Optimization in the advertising world should instead be thought of as the process where you find the most efficient way to spend your advertising budget with the end goal being to maximize your return on investment, whatever that may be. You can still optimize towards whatever you want (clicks, engagement, conversions, revenue, etc) without needing to define an explicit mathematical model. The process involves running tests, analyzing your results, and refining the contributing variables.

Why should you automate your optimization?

Remember, time is money. The market is getting more competitive as advertisers are getting smarter and understanding the importance of segmentation and A/B testing. This means more Ads, Ad Sets and Campaigns to create, analyze and manage for you; and more to compete against for delivery on the market. To get on top of the game you need to out-smart your competitors; this means spending less time on autonomous labour and more time on applying your learnings to make smart, strategic decisions.

To optimize your social media advertising, you want to:

  1. Segment your campaigns using the Facebook & Twitter best practices to set up a fair split test of your target audiences and creatives
  2. Find the type of people who are responding the most in the way that you want
  3. Find which type of creative these people are responding to the most
  4. Stop or reduce spending on poorly performing audiences and/or creatives and allocate your budget towards your highest performing ad sets or campaigns
  5. Refine your bidding so that you are paying the right price for each audience (i.e. only bid what that audience is worth to your business)

Refining each factor is a two-part process. First, manually dive into your campaign data to find which audiences and creatives are best working for your objectives; this is easily achieved using our Actionable Reporting tool. Then, make adjustments to your campaigns (i.e. budget, bid, status {on/off} etc) based on your findings from the first step. This second step can be either a manual or an automated process, using our flexible Automation Assistant to configure logic that lets the system make changes on your behalf. Reducing the manual labour in making these adjustments allows you to save time and effort, and helps reduce human error.

Case Study: Check out how Molson Coors used the Automation Assistant to automatically turn their weather-specific Ads on and off depending on the current weather conditions to optimize engagement.

How do you use the Automation Assistant?

The Automation Assistant is a fully configurable, highly flexible tool that operates across both Facebook and Twitter in the AdParlor system. There is no black box magic – users have full visibility and control of each action the system performs on your behalf. Briefly, the Automation Assistant allows you to configure Rule Sets; these contain individual Rules that can be executed in order. Each Rule itself operates on an “if this, then do that” basis.

We recommend grouping your Rules together either by functionality (what you want to achieve) or by the entities they are being applied to (Campaign, Ad Set/Ad Group, Ad/Tweet). Then, start building your Rules:

  1. Pick the scope of your Rule
    (e.g. this rule will apply to all my Facebook Ad Sets within Campaign X)
  2. Choose the time frame of data you want the Rule to analyze
    (e.g. lifetime, moving window, fixed dates)
  3. Set your ‘if’ condition
    (a condition is either a piece of campaign data e.g. CPA, CVR, clicks etc, or a factor like time of day or current weather conditions)
  4. Set your ‘action’ to perform when that ‘if’ condition is met
    (available actions depend on the scope selected in step 1 e.g. if you selected a Facebook Ad Set then you can adjust anything controlled at the Ad Set level such as budget or status)
  5. Choose how often you want the ‘if’ condition to be checked
    (e.g. I want to update my Facebook Ad Set budgets every 24 hours)

Common uses of the Automation Assistant

  1. Pausing Ads which are receiving delivery but are not generating any conversions
    i.e. I want to stop spending money on poor performing Ads so I have more remaining budget to assign to my top performing ones
  2. Increasing my budget for Ad Sets which have top performing Ads within them
    i.e. I have decided that “top performing” means my conversions are coming at or below my target CPA of $3 USD. Your criteria for “top performing” will depend on your business goals.
  3. Decreasing my bid values for Ads which are not generating a ‘satisfactory’ CVR%
    i.e. my Ads are receiving delivery but I’m not generating as many conversions as I’d like; I want to reduce the amount I’m spending on each of my Ads with poor CVR% without dropping out of the market altogether
  4. Sending a notification to a user when the conversion rate (CVR%) for one of my Campaigns has fallen below my ‘satisfactory’ threshold
    i.e. this will notify me so I can quickly troubleshoot the issue
  5. Turn on/off my weather-specific Ad Sets depending on the current weather conditions

Best Practices

Bid/Budget adjustment: Facebook recommends you only make up to 3 bid and budget adjustments per day to ensure their delivery pacing algorithms are not affected.

Rule frequency: This obviously varies depending on the scale of delivery achieved, but we typically recommend checking bid and budget rules on a daily basis (24 hour frequency) or less than 3 times per day. Weather-dependent rules should be executed more frequently (15-30 minutes) as weather data changes much more frequently.

Rule design: It’s best to decide on your logic framework before you begin implementing the Rules configuration; make sure you have all edge cases covered. Remember that new Ads will have no data; make sure these Ads have had sufficient delivery to be able to make a decision based on significant data (adding an extra condition which only lets the rule execute if more than some number (e.g. 1000) of impressions of the Ad have been served is a good way to do this).

Rule order: Rules are executed in descending order within a Rule Set, i.e. Rule 1 is executed first, followed by Rule 2 and so on. This means that Rule 2 will override the effects of Rule 1 if they are scoped to the same entity and conditions (although if you have configured your logic correctly this should never occur).

Using the does match IF and does not match IF action:

When the ‘if’ condition is not met, then no actions will be performed at all. When the ‘if’ condition is met, then the does match IF and does not match IF logic is applied according to the following logic:

if (condition is met for at least one of the entities at your chosen scope) {
for each entity at your chosen scope { // e.g. Ad Set within Campaign X
if (entity meets condition) { // i.e. does match IF
do something
}
else { // i.e. does not match IF
do something else
}
}
}
else {
do nothing
}

Get Started with AdParlor

Want to know more or get an account setup? Contact our support team at support@adparlor.com.