By Will Conboy
How do you ensure that your Facebook ad campaign is working as hard as it possibly can?
For Facebook advertisers, it can be easy to get stuck into the multitude of statistics and metrics that your campaigns generate.
For those that are analytically minded, Facebook marketing can feel like being a kid in a candy shop, but can be equally daunting to the uninitiated.
Reach, the number of unique individuals have seen your post, gives you an indication of just how far your ad has been seen, but it doesn’t really give you much of an indication of how effective it is.
It’s a similar story for impressions, which tells you how many times your ad has been served. The key metric is the latter divided by the former – frequency.
Frequency is a measure of how many times a user has been exposed to your post, and acts as an indication of how effective your targeting is. If your frequency is low, you are arguably targeting too wide an audience, investing too little to reach your target audience, or a combination of the two.
There are different ways of viewing frequency, and Facebook divides your campaigns into three levels; you can view the frequency for each ad, each group of ads (an ad-set) or each campaign (a collection of ad-sets).
Ad level frequency
Ad level frequency is the simplest level of frequency analysis. In the example below, each dot refers to a unique user, with the encompassing circle representing one ad.
As users who have already seen the ad are served it again, the impressions increase. However, because the number of users hasn’t increased, so does the frequency (because a greater number of impressions are served to a static number of users).
In this example, we have 19 unique users (represented as individual dots), with an ad that has been served a total of 27 times (once to 13 people, twice to four people and three times to two users). This generates a frequency of 1.42.
Ad-set level frequency
Ad-sets are designed to group ads together, making it easier for advertisers to organise and manage multiple campaigns, and manage their collective spend and targeting. This allows the targeting for many ads to be changed quickly and efficiently.
That might be great for campaign management, but it’s not so brilliant for frequency management. Ad-sets track all of the users who have visited all ads and if an individual is served several ads from an ad-set, that will still only count as one unique user.
One user that sees five ads would be considered as “one unit” of reach, five units of impressions and so that individual’s frequency would be classed as ‘five’.
In this instance everything within the larger oval is the ad-set.
- The green Ad still retains its 19 unique visitors, 27 impressions, and Frequency of 42
- The red Ad has 20 unique visitors, 22 impressions (as 2 individuals see the ad twice), giving a Frequency of 1
- 12 of the users however, have seen both Ads, with 1 user seeing both Ads twice
- The total unique users is 27 (not 39, which would be a combination of both ads)
- The total ad-set impressions is 49 (which is the combination of both ads)
- This pushes the frequency of the ad set up to 81, despite the green ad having a frequency of 1.42, and the red ad having a frequency of 1.1.
As both ads in this ad-set are served with the same targeting, they’re just as likely to be served either ad. Adding new ads within ad-sets would only compound the problem, so you should look at limiting ad volume within ad sets wherever feasible.
Campaign level frequency
A cursory look at the next level of frequency reporting potentially complicates matters further. In campaign level frequency analysis, we have a situation where a campaign may include several ad-sets, each with their own targeting.
That’s not really an issue when each ad-set includes mutually exclusive targeting, typically when negation is applied, but as we can see from the example below, many targeting approaches will overlap. After all, your audiences may have many interests.
A user may be a fan of a specific confectionary brand page already, or they may not – they can’t be both. However, complications arise when the targeting may seem mutually exclusive at first glance, but in actuality it isn’t. A fan of chocolate may also be a fan of candy, which means that they are likely to see ads from both ad sets.
In the above example, ad-set one (to the left) has two ads, as does ad-set two (to the right). Ads one and two, in ad-set one, have 31 unique users between them. As some individuals have seen each ad several times, their impressions is at 71, giving that ad-set a frequency of 2.09.
Ad-set two (to the right) has 35 unique users, and 82 total impressions, giving that ad-set a frequency of 2.34.
However, some of those users will feature in both ad sets, so the campaign as a whole has 54 unique users, and 153 impressions. That’s a frequency of 2.8, which is a step up from both ad-sets.
So how does this work in practice?
Let’s take an example of the sweet company wanting to promote their latest line of confectionary products.
Of the two products that they want to promote, one includes gelatine based products, and the other contains chocolate products, the brand therefore creates a campaign with two distinct ad-sets: one aimed at gelatine lovers, and one at chocolate lovers.
One of our target audiences, Tom loves gelatine, whilst another member of audience, Dick, loves chocolate. However, poor Harry loves Turkish Delight. Unlike Tom and Dick, he gets targeted by both the gelatine product ads and the chocolate product ads. While Harry is technically well served by the targeting, he gets an ad about milk chocolate, then an ad about jelly beans, rather than just one type of confection.
That may not sound like a huge problem, but it doesn’t give you a true indication of your Facebook ad strategy.
So what level should you listen to?
Campaign level frequency will provide you with the truest picture, but that alone won’t tell you the full story behind your campaigns.
Check your campaign. If it’s got a low frequency, then no problem. If it’s starting to creep up, then look at the ad-set level. If the ad-set levels are fine, then the first thing to look at would be to see if and where your targeting is duplicated.
If one or more of your ad-sets are high in frequency, then look at your ads. See if you have too many, turn them off, review how broad or narrow your targeting is, and look to either expand, or turn the ad-set off completely.
Why is high frequency an issue?
With the average internet user exposed to 1,707 banners each month (Comscore), campaigns can be serving your ads to people who just aren’t paying attention to them. They come up in their feeds, but they don’t get seen or actioned upon, leading to what is sometimes known as “banner blindness”.
Banner blindness (individuals learning to sub-consciously recognise, and then ignore your Ads), and campaign fatigue (signs of the campaign being less effective as time goes on) can be disastrous for your ads campaigns. It leads to soaring costs and diminished returns, but this isn’t the worst case scenario for a campaign.
If users receive several iterations of an ad over and over again it can start to feel like spam, and breeds negative sentiment towards the brand. Even worse, given that the ads are being displayed on a social platform, it’s easier for users to vent their frustrations as comments that remain visible to all future audiences.
So just what should your frequency be?
There’s a fine line, and much debate, on whether or not individuals will get annoyed, or whether they’ll see an ad enough times to be finally convinced to click.
The size and type of targeting that you’re doing will ultimately determine what the optimum frequency should be.
Smaller campaigns, focusing on a smaller number of individuals, will naturally reach a higher frequency, but if the ads are more accurately targeted, a higher frequency might be less of an issue.
If you focus on an audience that loves bear shaped gelatine products in London only, but the ads are tailored to the London Gelatine Bear Emporium just around the corner from where the audience lived, they may take more kindly to seeing the ads more often.
While it’s less likely to reach a higher frequency as quickly, broader targeting may start to annoy users who see the ads multiple times.
Users may become much more sensitive to how often they’re seeing the ads, given the drop in relevance with the targeting. Non-specific ad copy is for anyone, and being told about sweets 11 or 12 times when you’ve seen the message, taken it on board, and decided to do something about it (even if that decision is to ignore it) the message serves only to frustrate an audience.
As a rule of thumb, a frequency between five and ten for an entire campaign is still acceptable, but you have to review this at a campaign level to ensure it encompasses every individual that sees every ad, and not just focussing on either ad set one, or ad set two.
Splitting the report up will count a unique user every time they’re served an ad, giving false duplicates. Only viewing at campaign level with no other splits (demographic/day by day/device etc.) will show the true story.
So how do you avoid ad frequency mayhem?
Having established that there is no defined “right or wrong” answer to the frequency conundrum, how do we go about ensuring that we can at least control our ad campaigns? There are some key stages to go through:
Do your targeting homework
Understand your audience and use some of the extensive tools that are available to you in order to find them. The more you know about your audience, the better your targeting can be.
Plan your targeting coexistence
Could there be any cross-over between your ad-sets? If so, try to separate one group from the other to minimise the risk of duplication.
Choose your budget wisely
If an audience is particularly niche, track the reach, and campaign duration. If you plan to reach 100 people a day, are expecting £1 cost-per-click and the campaign lasts seven days, then £700 for the campaign will need every single user to click on the ad (or for some users to double click). Calculate your possible CTR and if it’s unrealistic, plan ahead by reducing spend or widening your targeting pool.
Manage campaigns, ad sets, and ads properly
Don’t setup several campaigns to target specific groups of people that will likely contain similar users. For example a group that likes chocolate, and a group that likes Mars Bars – one of these groups will largely encompass the other. Frequency won’t show up at an ad-set level report as Facebook will tally each individually, and your folly will only appear under the campaign level reporting.
Know how to report on frequency
Look at the big picture. Don’t review frequency on a day by day basis, ad by ad, or even ad-set by ad-set (each splits up your unique users). Look at the whole campaign.
Limit ad volume
A large volume of ads pushes Facebook to serve as many of them as possible to the same audience, increasing the likelihood of frequency jumping high.
While you may be tempted to include many ads to A/B test your creatives, don’t do this to an excessive level. A couple of ads per ad-set should be your limit. Trust your creative!
Review your ad relevance score
If some of your ads have a low score, it’s likely not resonating with the audience, so remove it. This removes the ad from collecting additional reach.
Will Conboy is head of marketing communications at Stickyeyes and a contributor to SEW. This article was co-written with Jonathan Hemingway.
Source:: Search Engine Watch RSS