The way things were…

I start­ed upload­ing videos to YouTube a decade ago. Back in August 2006, we only had a cou­ple of met­rics for mea­sur­ing our suc­cess. One was views, which count­ed the num­ber of times that our video was watched.

The oth­er was rat­ings, which showed the num­ber of stars that a small per­cent­age of view­ers had giv­en our video. Five stars meant they loved it; one star meant they loathed it.

In Sep­tem­ber 2009, YouTube replaced the five-star rat­ing sys­tem. Why? Well, as the chart below illus­trates, the over­whelm­ing major­i­ty of videos on YouTube had a stel­lar five-star rat­ing, while only a few view­ers were moved to rate videos when they did­n’t like them.

In oth­er words, peo­ple either liked the video they’d just watched or they couldn’t be both­ered to rate it. That made the five-star rat­ing sys­tem pret­ty use­less. And short­ly after that, YouTube replaced it with the thumbs up, thumbs down rat­ings that Face­book had made pop­u­lar.

And the way things are now….

Today, it’s time to replace the view met­ric.

Why? Because it has also become pret­ty use­less. In fact, there are the three key rea­sons why video ana­lyt­ics needs to fun­da­men­tal­ly change… again.

1. A “view” is not a view.

We’re now liv­ing in a mul­ti-plat­form video world where a “view”, is not a view. YouTube says you pay for a “view” when a view­er watch­es 30 sec­onds of your video – or the dura­tion if it’s short­er than 30 sec­onds – or engages with your video, whichev­er comes first.

On the oth­er hand, Face­book says you pay for a “view” when a video is dis­played in a user’s news feed for 3 sec­onds or more, even if the per­son doesn’t actu­al­ly click on the video to watch with the sound turned on.

So, with­out a stan­dard­ized def­i­n­i­tion of “view,” com­par­ing the num­ber views that your YouTube videos have with the num­ber of views that your Face­book videos have is like com­par­ing oranges and kumquats.

2. A view has no quantifiable value.

It’s time to admit that no one real­ly knows what a view is actu­al­ly worth. How many views do you need to increase brand aware­ness? How many views does a video have to get to have a direct impact on con­sumer per­cep­tions and behav­iors?

So, if a view isn’t a met­ric that mat­ters, then why would any video mar­keter want to con­tin­ue using it as a key per­for­mance indi­ca­tor (KPI)? That’s not a KPI that will earn you a pro­mo­tion or cost-jus­ti­fy a big­ger bud­get.

3. As things stand, it’s impossible to budget based on view.

Third, if no one knows what a view is actu­al­ly worth, then what is a good aver­age cost-per-view (CPV)? Well, it’s prob­a­bly going to be as low as pos­si­ble, which isn’t one of the health­i­est trends in the dig­i­tal video mar­ket­ing busi­ness.

This is bad for the plat­forms like YouTube and Face­book, which are sell­ing video adver­tis­ing. It’s espe­cial­ly bad for the YouTube con­tent cre­ators who are try­ing to mon­e­tize their videos with CPV adver­tis­ing. And low-cost CPVs are not even par­tic­u­lar­ly use­ful replace­ment met­rics for the ad agen­cies that are still using gross rat­ing points (GRPs) to mea­sure their TV cam­paigns – espe­cial­ly with more and more of their clients are ask­ing unan­swer­able ques­tions like, “How many GRPs do I need to sell a car?”

So, what are the alter­na­tives? For­tu­nate­ly, there are now some good con­tenders to replace views as a met­ric.

Three alternative metrics to measure video marketing success

1. Watch time

First, video mar­keters should replace “views” with “watch time.” Why?

Well, about a year ago, the Google Agency Blog pub­lished some research which found that there is a con­sis­tent rela­tion­ship between how long a video ad is view­able and increas­es in brand aware­ness and con­sid­er­a­tion. The longer a user views your ad, the high­er the lift in these two impor­tant brand met­rics.

Watch time is also a met­ric that mat­ters for the two lead­ing video plat­forms.

In Octo­ber 2012, YouTube added “esti­mat­ed min­utes watched” to the “views” report in YouTube Ana­lyt­ics. A day lat­er, the video shar­ing plat­form start­ed adjust­ing the algo­rithm for the YouTube search results, ben­e­fit­ting engag­ing videos that kept view­ers watch­ing.

In June 2014, Face­book announced that it was adding how long some­one watched a video to its rank­ing fac­tors, which had pre­vi­ous­ly includ­ed likes, com­ments, and shares. And in Feb­ru­ary 2016, Face­book added new met­rics to its reports that let adver­tis­ers see the per­cent­age of peo­ple who have viewed their videos with sound in Page Insights and Ad Insights, too.

2. ‘Brand Lift’

Sec­ond, video mar­keters should replace their assump­tions about the puta­tive val­ue of video “views” with actu­al mea­sures of “brand lift” or the rapid results of “brand polling.” How?

In Feb­ru­ary 2013, Google intro­duced its Brand Lift solu­tion. It ini­tial­ly mea­sured mea­sure the actu­al impact of a YouTube ad cam­paign on brand aware­ness, ad recall and brand inter­est. Today, it also mea­sures the impact of YouTube ads on con­sid­er­a­tion, favor­a­bil­i­ty, and pur­chase intent.

How does Google’s Brand Lift work?

Google’s Brand Lift solu­tion mea­sures the above in two dif­fer­ent ways:

To mea­sure brand aware­ness, ad recall, con­sid­er­a­tion, favor­a­bil­i­ty, and pur­chase intent, Google iso­lates a ran­dom­ized con­trol group that is not shown your YouTube ad, and an exposed group that does see your ad. About a day after see­ing (or not see­ing) your ad, Google Con­sumer Sur­veys deliv­ers a sur­vey to both groups. Since the only effec­tive dif­fer­ence between the two groups is whether they saw your ad, Google can accu­rate­ly deter­mine the lift attrib­uted to your cam­paign.

Google’s Brand Lift solu­tion also mea­sures the impact your YouTube ad cam­paign has on cre­at­ing inter­est in your brand by using organ­ic search­es on both Google.com, the world’s largest search engine, and YouTube.com, the world’s sec­ond largest search engine. Sim­i­lar to sur­veys, Google ran­dom­ly picks a group that saw your ad and a con­trol group that did­n’t see your ad. Google then com­pares the organ­ic search behav­ior of both groups, look­ing at how often they search for key­words relat­ed to your brand or cam­paign. The dif­fer­ence in search­es can be attrib­uted to your cam­paign.

After ana­lyz­ing around 50 cam­paigns from For­tune 100 brands and cat­e­go­ry lead­ers run­ning on Google Pre­ferred, which includes some of YouTube’s most pop­u­lar chan­nels, Google found that:

• 94% of the cam­paigns drove a sig­nif­i­cant lift – an aver­age of 80% – in ad recall.
• The analy­sis also found that 65% of Google Pre­ferred ads saw an increase in brand aware­ness, with an aver­age lift of 17%. This is par­tic­u­lar­ly impres­sive con­sid­er­ing that the brands in the study were already well-known.

Here’s a video that explains how Google’s Brand Lift solu­tion works. It also men­tions how Mon­delēz Inter­na­tion­al used Brand Lift to mea­sure the mar­ket­ing effec­tive­ness of its Tri­dent Unlim­it­ed launch in Brazil, and Nis­san Cana­da used Brand Lift to mea­sure the effec­tive­ness of True­View ads for its Micra mod­el.

How does Facebook’s Brand Polling work?

Face­book and Insta­gram also offer brand polling solu­tions that enable adver­tis­ers to mea­sure brand aware­ness and ad recall. Just like Google, Face­book ran­dom­ly puts part of your tar­get audi­ence into a con­trol group that isn’t able to see your ads. The poll is then served to the con­trol group and a test group with iden­ti­cal char­ac­ter­is­tics. The lift between the con­trol and test groups accu­rate­ly shows the impact of Face­book and Insta­gram adver­tis­ing.

3. Cost-per-acquisition (CPA)

Third­ly, adver­tis­ers should move beyond opti­miz­ing their video cam­paigns to get the low­est CPV and start using cost-per-acqui­si­tion (CPA) bid­ding for video ads instead.

For a YouTube ad cam­paign, you can use a CPA bid strat­e­gy to tell AdWords the amount you’re will­ing to pay for a con­ver­sion. AdWords will then auto­mat­i­cal­ly sets your bids to give you as many con­ver­sions as pos­si­ble at your tar­get­ed cost-per-acqui­si­tion.

This strat­e­gy can help you reach cus­tomers who are like­ly to take an action, such as mak­ing a pur­chase (for an ecom­merce site), com­plet­ing a game lev­el (for a mobile gam­ing app), or sub­mit­ting a con­tact infor­ma­tion form (for a mar­ket­ing or lead gen­er­a­tion site).

Tar­get­ed CPA bid­ding focus­es on max­i­miz­ing con­ver­sions for your video ads, rather than views, as in CPV bid­ding. You still pay per view, but AdWords auto­mat­i­cal­ly sets your bids to help you get more con­ver­sions.

Case Study: Gillette BODY

For exam­ple, Gillette tar­get­ed a rapid­ly expand­ing audi­ence of body-groom­ing men to cre­ate buzz around its lat­est prod­uct, Gillette BODY. With a cam­paign anchored by a YouTube ad enti­tled “100 Years of Hair” the brand got more than 14 mil­lion views (if you still want to count those!).

The stats were far more impor­tant:

• 84% of view­ers watched at least 85% of the ad.
• Gillette search­es grew by 211% and clicks grew by 111% (biggest increase ever for the brand)
• And the cam­paign deliv­ered more than 500,000 clicks-to-buy Gillette BODY, sur­pass­ing sales expec­ta­tions by up to four times across sev­en mar­kets.

Face­book also helps adver­tis­ers deter­mine exact­ly how their video ads impact their bot­tom line by offer­ing con­ver­sion lift mea­sure­ment capa­bil­i­ties. Build­ing on exist­ing Face­book mea­sure­ment offer­ings, con­ver­sion lift allows adver­tis­ers to accu­rate­ly deter­mine the addi­tion­al busi­ness dri­ven by Face­book ads and make future mar­ket­ing deci­sions based on this infor­ma­tion.

Case Study: Open Colleges, Australia

For exam­ple, Open Col­leges, Australia’s lead­ing online learn­ing provider, used a con­ver­sion lift study to mea­sure the full impact of its Face­book cam­paigns. Results from the study revealed:

• a 95% lift in con­ver­sion rate for web­site inquiries
• a 12% lift in offline enroll­ments.
• cost per acqui­si­tion was also 23% low­er than what Open Col­leges’ last click mod­el had indi­cat­ed.

Summary: What does this mean for video analytics?

In short, the video ana­lyt­ics that you’ve been using for the past decade need to fun­da­men­tal­ly change if you want to mea­sure what mat­ters today. Imag­ine being able to go beyond tra­di­tion­al met­rics like views to mea­sure your video ad campaign’s impact on brand aware­ness and con­sid­er­a­tion.

Imag­ine being able to answer tough ques­tions from skep­ti­cal exec­u­tives who want to know:

• Do peo­ple truth­ful­ly recall watch­ing our video ad?
• Are our tar­get con­sumers real­ly more aware of our brand after view­ing our video ad?
• Did our video ad actu­al­ly move peo­ple to con­sid­er our brand or prod­uct?
• Are con­sumers even more favor­ably aligned with our brand’s mes­sage or iden­ti­ty after view­ing our ad?
• Are con­sumers tru­ly intend­ing to pur­chase our prod­uct after see­ing our ad?
• Are peo­ple gen­uine­ly more inter­est­ed in our brand or prod­uct?
• Did our ad hon­est­ly increase organ­ic search activ­i­ty for our brand or prod­uct?

And final­ly, just imag­ine the impact on your career as well as your company’s suc­cess if you begin to focus your video ad cam­paigns on the high­est val­ue con­ver­sions with the low­est pur­chase amount.

That’s a pow­er­ful rea­son to unlearn what you have learned back in 2006. And, that’s why video ana­lyt­ics needs to fun­da­men­tal­ly change today, not 10 years from now.

What are your thoughts on how we should be mea­sur­ing video mar­ket­ing suc­cess?