Intro

How can success be measured for an online brand campaign? With YouTube as one of the main online channels, we used Google’s Brand Lift to measure the impact on the brand of CM – Belgium’s largest public health insurance company – during their 2016 Brand Campaign.

Different from CM’s ongoing performance campaigns throughout the year (where conversions such as new leads are the main objectives), we were looking for a way to measure the impact and thus the success of the branding campaigns. Needless to say we were very excited Google came up with their Brand Lift surveys and we we could finally implement it for one of our clients.

First time you hear about Google Brand Lift and wondering why they came up with it? Read this 2015 post on Think With Google and you’re good to go.

So how does it work?

The basic idea is as simple as it sounds. Before take-off Google, divides your audience randomly in two groups:

  • The Control Group: this is the group of people who did not get to see your ad on YouTube.
  • The Ad Exposed Group is the group of people to whom your video ad was shown (and hopefully remembered it).

About a day after seeing (or not seeing) the ad, a brief survey is delivered to both of the groups. “Since the only effective difference between the two groups is whether they saw your ad, the attributed lift to the campaigns can be determined accurately”, Google says. The survey pops up in YouTube instead of another pre-roll ad, and there is no obligation in answering. Halfway our campaign we had a survey completion rate of 15%.

Brand Lift survey

 

Google checks the impact of your campaigns on your brand on three levels:

  • Ad Recall: Do people remember seeing an ad of CM?
  • Brand Awareness: Is CM (more) top of mind within our “Ad Exposed Group”?
  • Brand Interest: Is there an increase in search frequency on brand related search terms?

Ideally you would see a positive impact on all levels within the “Exposed group”.

First step in setting up Brand Lift: Get in touch with your Google representative

That’s right: you can’t set up the surveys yourself. If you want to start with Brand Lift you need to reach out to the Sales representative within Google who is supporting your account(s). The main reason for this is that they want to assure that everything is implemented in a proper way.

Brand Lift Booking Request Google

Since there is a threshold of 20k spend (with a minimum spend of 7k spread equally in the first 7 days of the campaigns), no chances can be taken to implement the surveys in a wrong way. An incomplete or incorrect setup would most likely result in false information and analysis afterwards. So Google takes care of the survey forms based on our input, and the advertiser or agency in this case takes care of the campaigns.

Using the Brand Lift Booking Request you can submit your application. Please note that within the General section, point 3, they are talking about a minimum spend of €1.000 for the first 14 days but this was set to 7 days in our case. Nonetheless, the minimum spend overall can’t be less than 20k. Beware that when one of the thresholds is not reached, Google will stop the surveys immediately.

Requirements & settings

So Google arranges the survey-side of the Brand Lift, but you still have to take care of the setup and management of the YouTube video campaigns within the Adwords Account. It’s not much different from setting up “regular” video campaigns, but there are some things you have to keep in mind:

  • New campaigns:It’s not possible to use or recycle old campaigns who ran previously. No historic data is allowed in the campaigns.
  • Fix start date: All video campaigns should have the same scheduled start date & time.
  • Campaign status: Should be Paused when setting up. Active campaigns can not be linked to Brand Lift surveys, even if there is no data in them yet.
  • Maximum of 5: A maximum of 5 campaigns is allowed. In our case we used 3 campaigns, one for each theme (“Welkom”, “Betaalbaar”, “Meegroeien”).
  • Targeting: Don’t target too narrow so you are sure the minimum spend will be reached the first 7 days.
  • Video ads: Preferably you work with only one base video ad, so the same visual & message is spread everywhere.

With all people seeing the same video ad this should increase the chance of a positive lift on your brand. Since we wanted to optimize the campaigns mid-flight, we did work with a variation in the ad to see which ads generates the highest number of complete views.

As said previously, we chose to build a campaign for each of the three themes of the branding campaign. Since CM serves people of all ages in all regions in Belgium we did not target any specific topics, interests or placements.
On the ad group level, our segmentation was based upon age and gender so we could optimize efficiently as the results came in, based on their behaviour. For a brand campaign (especially when in combination with Google Brand lift) within YouTube, it’s recommended to optimize on VTR (up to 100%) and not on CTR’s or even conversions (if any). We want to make sure that as many people as possible watch our video ads completely, which is 17 seconds in our case, and limit the number of people who drop out after the Skip Now or perhaps halfway the ad.

This philosophy is supported by Avinash Kaushik’s “See, Think, Do, Care” model, where video views is one of the metrics in the “See”-fase of the framework to measure brand awareness.

Launch, cross fingers and wait for first results coming in

Campaigns were launched on the 14th of April 2016 and scheduled to run until the 11th of May, which comes down to a total run time of about 4 weeks. First results of the surveys were expected to come in after about 10 days which was also the case here. You can not monitor the results yourself so you depend on the information given to you by your contact at Google.

Unfortunately, there is no significant impact on the brand so far according to Google’s first results. Are we disappointed? Maybe a bit, because the expectations were high at the start. On the other hand, we are still only half way and these first insights might give us the opportunity to improve the campaigns even more.

It is likely to assume that the impact (either positive or negative) on our brand is determined by people who have seen the ad completely. Therefore we are aiming to have as many people possible with a Video Played To Rate of 100%. The result of our halfway actions are visible in the below graph, where you can see that the View Rate increases week after week, and the cost per view drops from 0,07€ to 0,03€. If you’re curious on how we did this feel free to contact us.

Graph for Google Brand Lift

We mentioned earlier that about 15% of the people who were questioned about CM’s ads submitted their answer. Knowing this number (N=) is 7,252 we can tell that over 48,000 people were asked to participate until now. That’s a lot! Imagine running a traditional campaign on TV for which you want to measure the impact afterwards. Needless to say it would take you ages (and money) to question the same amount of people on the street or by phone.

Final results are finally here

Final results Google Brand Lift

About a week after the end of our campaigns we received the final results in a slide deck. Lots of information and print screens but unfortunately one thing is clear: there is no lift in Brand Awareness and Ad Recall. Not overall and not when we look at the results more in depth (on a level such as gender, age, device, geo, frequency, campaign,..).

When we take a closer look to the results by age for example we can see the exposed group (dark blue) is outperforming the control group until the age of 35-44. Hereafter it’s the other way around. The impact is too little (take a look at the overall bars) to talk about any lift.

But there is some good news as well! On the 3rd field (Brand Interest) there is a positive impact noticeable on the brand. Briefly this means that the brand-related search behaviour of the people within the exposed group in the days after seeing the video ad(s) have increased significantly enough to talk about lift:

Taking a closer look at the geographical regions we see a big difference between the top 4 regions compared to the overall average:

So yes, we’re glad there is an impact on the brand when it comes to search behaviour but it disappointed me that no lift was generated with the Recall / Awareness part. What could be the reason for that, because normally the 3 legs of Brand Lift should go hand in hand?

CM campaign ad example

One of the main remarks we should make is that the YouTube campaigns were part of a multi-channel strategy. So besides running the video ads on YouTube nearly the same ads (edited to 8”) ran within Facebook and Instagram. Next to that the characters from the videos were also used in display banners running in the same channels + GDN. Combining this with airing the ads during prime-time on national tv it makes sense thinking this could impact the so-called control group. Within YouTube they are “marked” as people who did not get to see the ad but it’s realistic to think a part of them saw the ad on tv, which will affect them answering the surveys within YouTube.

Next to that it’s also possible that you as a user are logged in to YouTube on your desktop, and not on your mobile device. So theoretically it could occur that you are (without knowing of course) part of the control group on one device, and having the ad exposed on the other device.

Would we do it again for this customer? Most likely yes, but not in the same set-up as we did here. The results and insights can be very valuable but only if we can assure the data is not polluted or influenced by other channels, campaigns or visuals.

About CM

The Christelijke Mutualiteit (better known as “CM”) is Belgium’s largest public health insurer. With a market share of circa 45%, serving over 4,5 million clients nationwide, they are by far the leader in their branch (#2 follows at ca. 27%). Every other year they launch a national branding-campaign to promote their products and even more important; their services.

Being their digital partner, BossData is in charge for the set-up, execution and management of the online part of these brand awareness campaigns.
Traditionally for these type of campaigns most of the efforts and therefore the majority of the budget was allocated to offline channels (direct mails, TV- & radio-commercials, etc). However, since a couple of years they are convinced that the role & importance of online channels should not be underestimated. It’s not only about the power of reaching the exact audiences that we want. Being able to measure the outcome of our actions and act on them mid-flight is without a doubt the biggest benefit for these kind of campaigns. Which doesn’t automatically mean there isn’t any budget for TV anymore.