Good morning, I am delighted to be back in Mexico City once more. I am Chief Global Analyst at Millward Brown and, as such, come from a somewhat different world from many in this room. Millward Brown works with many of the world ’ s largest brands, helping them to build their brand, drive sales and improve brand value. For us, online and digital is an important means for companies to build their brands but it is not the only one. We devote considerable time and resource to ensure that we can measure the full effect of digital media spend and understand how it fits with other more traditional media. In my presentation today I wish to review some learning on what works to build brands online. Not to drive direct response but to build affection and the predisposition to buy a brand at a later date.
In the first section of my presentation I want to consider the metric that continues to dominate the lives of online marketers: click through. What it is, what it is not and how much it might really be worth. Click though has been referred to as “ the core of our digital nervous system, and the key to the online economic system ” by Ted McConnell is exec VP-digital for the Advertising Research Foundation (more from Ted later).
It is largely as a result of click through (and other behavioral metrics like expansion rate and video completion rate) that many people believe that online advertising is the most measurable media. Unfortunately, it is a belief I do not share. We can measure a limited number of metrics with complete accuracy online but these metrics only give us a partial, and in some cases misleading, view of our advertising ’ s effectiveness. The fundamental issue is that even though much of the advertising online is designed to affect awareness and attitudes we still measure its effectiveness as if it were a direct mail campaign. In a recent survey by Vizu of more than 450 advertising executives, marketers said that they will be allocating nearly 60% of their online budgets to brand advertising in 2012, indicating that spending on online branding may surpass spending on direct response advertising for the first time in the coming year. http://www.prweb.com/releases/2012/1/prweb9084906.htm As Millward Brown ’ s Chief Global Analyst it is my job to make sense of trends and numbers. So here is a number for you.
The number is 16. 16 years ago Millward Brown helped demonstrate the brand building power of online advertising for the first time. In 1996 we teamed up with a company called HotWired to measure the brand building effect of online advertising. The research was conducted as a response to a growing fixation in the Internet industry with a behavioral metric called click through. The proportion of people who click on an online ad. Our research, the first of its type, proved that online advertising could build brands just like any other medium. http://www.nytimes.com/1996/12/03/business/banner-ads-on-internet-attract-users.html Since then much has changed. We have sophisticated ad tracking and targeting systems. We have online video and rich media. We have social networking. But not everything has changed. The majority of the online advertising industry is still fixated with that simplistic behavioral metric of click through. My belief is that most digital advertising today is conducted without recognizing the bigger picture of how brands really make more money. So let me lay out my case by sharing a few more numbers.
In a recent survey 64% of advertising and agency executives claimed to use CTR to evaluate performance and optimize their online spend. http://www.collective.com/sites/default/files/Collective-2010-Display-Study.pdf A further 61% use Cost per Click and 48% cost per acquisition. In spite of the fact that the majority of online marketers claim to be using the medium for brand building only a minority used brand awareness or brand favorability to optimize spend. Why is the click so dominant? Because it is easy. Because it is universal. Because it is free. Unfortunately, simple metrics win every time. Yory Wurmser of the Direct Marketers Association gives us another reason. “ Transactional data tends to be more accurate, since it takes the human element out of the assessment and shows exactly what happened. ” Except, of course, it doesn ’ t give the whole picture. The problem with transactional data like clicks and likes is that it tells you what the minority of people did. It does not tell you why those people did something. And it does not tell you what the majority of people exposed to the ad did. Worse still there is plenty of evidence to suggest that click through is not accurate. But setting these pints aside I still find this statement bizarre. What on earth are we doing as brand or direct response marketers if not dealing with “ the human element ” ? Brands would not exist and certainly would not help businesses make more money without the human element.
I suggested that clicks represented a minority behavior. These days the number of people who respond to online advertising is very small indeed. According to DoubleClick 1 in a 1000 people actually clicked on the average ad in 2010. http://doubleclickadvertisers.blogspot.com/2011/08/latest-display-benchmarks-now-available.html That ’ s less than 5% of how many people clicked in 1998 when it was 1 in 50 people. It may be less than reported click through rates in Mexico but maybe it gives you a glimpse of the future. And wait, there ’ s worse news ahead. If you think that click through is an accurate measure of response to an ad think again.
Ted McConnell is exec VP-digital for the Advertising Research Foundation recently reported a little experiment he conducted in AdAge: http://adage.com/article/digital/incredible-click-rate/236233/ He and some associates ran six blank ads in three IAB standard sizes, and two colors, white and orange. The average click-through rate across half a million ads served was 0.08%, which would be good for a brand campaign, and so-so for a direct response campaign. When asked, half the clickers said they were curious, the other half admitted to a mistaken click. As McConnell admits the experiment is not realistic, after all it was a blank ad, but he states, “ At a minimum, the data suggest that if you think a click-through rate of 0.04% is an indication of anything in particular, you might be stone-cold wrong. ” So if half the people who click on an ad do so by mistake what does that imply about the value of online advertising? ROI calculations based on click through start to look even more dubious than they do now. And yet we know that the overall effect of an online ad is far greater than immediate behavior would imply.
In research conducted by Dynamic Logic over the last three years an additional 1.3% of people claim they will buy a brand as a result of being exposed to an online ad (compared to those exposed to a control ad). Source: MarketNorms database. That ’ s a lot higher than the click through rate. And it is interesting to note that the average for Latin America is the same for the overall database, suggesting that online in Latin America has great potential. Ah, you say, but that is intent it is not a purchase (but then neither is a click). It is behavior that matters. We need someone to buy the brand. So what proportion of these people with positive predisposition go on to buy the advertised brand? The answer is anywhere between 10% and 70%. There is a good relationship between overall purchase intent and sales in most categories at an overall level. The same is true when we look at the response to online ads specifically. Our research estimates that somewhere between 50-70% of people follow through on their purchase intent for consumer packaged goods after seeing an online ad. That ’ s a similar conversion to click through but the absolute number is much bigger. Estimates for automotive and entertainment are lower and in part depend on the purchase frequency. For CPG brands measured by SymphonyIRI the overall average ROI was $1.50 for every dollar spent. The truth is that few people are at all interested in thinking about brands when exposed to an online ad. Instead they want to get the news, find recipes, check out information on celebrities and connect with their friends. To think that the few that do click are in any way representative of the total is naïve.
Unfortunately all the evidence suggests that click through and purchase intent are not correlated. http://www.dynamiclogic.com/docs/presentations/2012/03/14/ARF2009_Behavioral_Branding_Correlation.pdf You cannot assume that a the total response to an ad is reflected in the click through rate. This was true of the ads we tested in the HotWired Ad Effectiveness Study and it still is. We just have far more proof. For instance, auto ads tend to get clicked on more than others but that does not mean that more people want to buy the advertised brand. Automotive ads tend to attract high click through rates but only average increases in purchase intent. Reason? There are plenty of people who are just interested in cars and want to know the latest about them even if they are not in the market to buy one. Financial services on the other hand score low on both click through and purchase shift. Reason? Few people are interested in banks and are unwilling to consider switching unless forced to do so or the offer is very attractive. The research from which this finding was taken was conducted in conjunction with Google in 2009 but recent work conducted by a start-up called Pretarget using comScore data finds the same thing for clicks and conversion. Pretarget found that the correlation between clicks and a conversion was 0.01. http://adage.com/article/digital/click-rates-matter-thought/234330/ So if you are optimizing your media buy against click through there is a good chance that you are either overestimating or underestimating the impact of the campaign…by a lot. Why does this matter?
Because almost 2 in 5 online ads measured by Dynamic Logic have a negative effect on purchase intent. Source: MarketNorms database. Did you really think that ads that pop-up or get in the way of people looking for content was really a good idea? Mmm… now what if people follow through on negative purchase intent as much as they do positive purchase intent? The implications of this finding are scary. Could a lot of online advertising actually be eroding brand value by undermining people ’ s predisposition to buy the advertised brand? I can ’ t answer that question today but what I can tell you is that by focusing on click through to the exception of other metrics we risk undermining the value of the very brands that pay online advertising ’ s bills. Let ’ s leave the numbers for a minute and ask ourselves what we are really measuring with click through.
What is a click? What does it actually measure? I would like to propose this definition. A click is an impulse expression of interest in a brand or offer. It is an action that implies that people not only find your ad of interest but that they have enough time and interest to check it out right now. More people may find your brand relevant but they just don ’ t have any time to follow up right now. More people may find your brand relevant but other things are more interesting to them right now. What I believe most online advertising does is act as an accelerant. A call to action that gets some people to do something now that they would have done anyway, given time. I believe that all too often the great ROI reported for online advertising, paid search and social advertising represents the penultimate step in a chain of events – including exposure to other forms of media – that leads to purchase. The question is would these people have bought anyway even without the call to action? But, of course, online advertising need not be limited to a call to action.
Last week my colleague John Svendsen, Global Brand Director for Media at Millward Brown, said this in conversation. “ All media are capable of doing all things on a per person basis. ” What he means is that with the right strategy and creativity any media can build awareness, create an emotional response or drive behavioral response. The only real difference between media channels is reach. All else is down to the advertiser ’ s mindset. There is absolutely no reason why online advertising cannot build brands just as effectively as other media. We proved that it has the capability to do this for 16 year.s And given the dramatic fall in click through since then one could argue that online needs to justify its existence by demonstrating that it can build value beyond being an accelerant of response.
So what do we need to ensure that online advertising builds brands effectively? We have seen a lot of change in the last 16 years and learned a lot of new things. But not everything has changed. In the remainder of the presentation I want to consider what remains true today and what is new.
The first thing that has stayed the same is that online advertising still has the capability to build brands.
Data from Ad Index studies conducted in Latin America prove this point. Exposure to online advertising still has the potential to raise ad and brand awareness and improve brand favorability and purchase intent. MarketNorms Q2/11 (Category: LATAM, N= 18 campaigns; n=19,060, last three years)
Importantly we still find that creativity matters to online advertising, just as it does with other media. The problem is that we often find that simple things that could improve the effectiveness of online are ignored. Online media is different from other, less actively consumed media. This means there several things unique to the medium.
This data was for an ad campaign tested here in Mexico. As you can see, the campaign delivered strong results. Ad awareness, the precursor to attitudinal and sales response rose nearly 28% between control and exposed cells. Purchase intent rose nearly 8%, over five times the average for Latin America. So why were the results so good? Because the brand followed the golden rules of online advertising.
Leverage the brands iconic assets and show the brand name prominently on all frames. You cannot guarantee that people will see all the frames of an ad. You can ’ t even be sure they will see the first one! Reveal ads that assume people will follow along to get the message are very likely to backfire. Keep your messaging simple. If you hope to create a positive response to your advertising do not use interruptive advertising. You only set up a negative reaction making it difficult for people to respond positively to any message.
In April 2001 I was here in Mexico City to present at another conference. This one hosted by ESOMAR, the international research organization. The title of my presentation was “ Is bigger really better? ” My answer to that question was yes. Bigger was better then and it still is today.
In 2012 we conducted a retrospective analysis of three new and three established online ad formats using Dynamic Logic ’ s MarketNorms ® data. While the results differed depending on the metric looked at the one that created the biggest bang for the advertising buck was the billboard. The large, full width ads that appear at the top of a page. Wallpaper ads, that essentially take up the full page but sit behind the content were next most effective. And whether sites and users like it this is the sort of advertising that we are likely to see more of. Because the first step on the road to brand and sales success is to get noticed. Source: Dynamic Logic Global analysis of 770 campaigns for IAB UK
Back in 2001 I also reported that streamed ads had a significant impact on brand metrics compared to the normal ad banner.
Streaming and online video have come a long way since then but our testing still demonstrates that video has more impact than other formats. Standard and rich media ads require substantially more exposures to match the first exposure impact of video. However, global data shows that with over-exposure, high impact formats can begin to drive down attitudes toward a brand, particularly in regards to purchase intent. http://mediatel.co.uk/newsline/2012/06/07/fast-forwarding-your-brand-with-online-video/
On the whole we ’ ve seen that in-stream ads are better for driving awareness and message association, but non in-stream may be better for driving favorability and consideration. Source: Dynamic Logic MarketNorms, Q4/09, Last 3 Years , In Stream Video N= 42 campaigns, n= 51,781 respondents; Not In Stream Video N= 209 campaigns, n= 234,949 respondents
So many of the basics of online advertising have stayed the same over the years. But many things have changed. The technology has evolved and we now have that vast new playground called social media. Over the years Millward Brown has sought to measure and understand the potential of these new opportunities.
One major change that marketers here in Mexico will have to come to terms with is that “ viral ” needs help.
I have to ask. Do you feel lucky?
We are perhaps less familiar with the use of the word “ organic ” when it comes to viral advertising but for a long time people believed that simply posting an ad online gave it the potential to go viral. If people discovered an ad they liked they would share it with others. Unfortunately, our research suggests that there is still no such thing as a free lunch. We have shown that ads that go viral possess similar creative properties to any other ad – only more so. They need to be extra creative to make people want to share them. Perhaps more importantly, we also find that viral advertising is not truly viral. In order to create good levels of pass along you need to publicize your content. For instance, in work conducted for YouTube we found that home page ads are directly or indirectly responsible for 86% of all views and improve overall viewership by a factor of 6 compared to unpublicized ads. Featuring the ad on the home page meant that even the ads with low viral potential got some views and only the ads with the best potential came close to matching the featured ad performance.
Back in the early days of online it was often proposed that the new medium would supplant all other forms of advertising. Instead it has created a sort of parallel universe. Online video is the equivalent of TV advertising. Banners are billboards. And web sites extended print ads. Moreover, it has become apparent that the Internet does not work alone. It has to be considered in the light of the overall media campaign.
The reason is simple. During the life of a campaign online advertising is almost always seen in conjunction with other media. In this example taken from one of our Cross Media studies for Nestle ’ s KitKat brand we found that less than 2% of respondents were likely to have been exposed to only the online display advertising. More people had actually seen the brand on Facebook.
But that is not to say that online – display or Facebook - did not play an important role. Our findings demonstrated that TV alone would not have created the same impact as the combination of media. TV had the biggest impact overall (reflecting its stronger reach) but Facebook had the strongest return on investment.
And mention of Facebook brings us to one of the biggest and most important changes in the last 16 years: the advent of social media. In 2011 87% of Mexican Internet users claimed to be registered to a social network. And most of them – 94% - claimed to be on Facebook (up from 86% the year before). Of these, 40% belonged to a fan page. And unfortunately that seems to be good enough for many marketers. All they seem to want is people to like their brand assuming that this will be noticed and encourage others to buy the brand. Heck, some marketers are even willing to pay for likes even though they could get them for free with the right strategy.
Unfortunately I believe that likes are very similar to clicks…only less so. In most cases they are superficial indications of brand allegiance and mostly given in response to incentives – offers of freebies and giveaways – or prior affection for the brand. For some time now we have been observing an arms race to get people to “ like ” brands on Facebook. The trouble is relatively few brands have figured out what to do with those people once they have liked the brand.
The first thing to do is to recognize that fans are important to you. They already outspend non-fans by over 4x. Source: Millward Brown ’ s BrandZ, 2010 global database. The fan page challenge is to deepen the relationship that you already have not try to pitch them or get them to persuade their friends to buy the brand. Given the right encouragement they will do that by example. What influences people to like a brand? Well, to start it depends on the country. Latin Americans are very positive about brands and tend to like them more easily than, say, those colder blooded Brits. Then it depends on the product category. Categories which encourage conversation; e.g. software/ hardware, diapers, telecomms, mobile phones and cars, tend to get more likes than others. And then brands which are particularly creative and desirable, and justify their premium, are likely to be liked. There is a strong correlation between brand equity and the number of likes on Facebook with the strongest brand getting the most likes. Challenger brands often pick up more than their fair share but otherwise most brands are liked by relatively few people.
In order to help advertisers maximize the impact of their Facebook page we provide a service called FanIndex. Exposed respondents are recruited directly from the fan page and o ne of the questions we ask is For each of the following attributes, has your opinion of BRAND changed since you became a fan and visited the fan page? In this case study for a brand here in Mexico we find that the Facebook page had a strong influence on perceptions that the brand offers something different from others. This is a really important finding because this perception helps define whether a brand can command a price premium or not. So what sorts of things help a brand to achieve this type of response?
In our analysis of effective Facebook pages four things stood out as differentiators: variety, interaction, useful information and community. http://millwardbrown.com/Libraries/MB_Articles_Downloads/2011_FanZ_Social_Media_and_Brand_Fans.sflb.ashx Here in Mexico… 79% want news about the brand. 51% want promotions (look out, it ’ s a trap!). 50% relevant information. 44% to know other users opinions. And 31% to communicate directly with the brand. ( Digital Media Consumption Study of Mexican Internet Users 2011 conducted by Millward Brown in conjunction with Televisa and IAB Mexico).
One of the major roles of advertising is to make a brand salient and vital. Salient in that the brand comes to mind easily and quickly when relevant. Vital in that people are drawn to brands that seem successful, growing, popular and contemporary. And increasingly people judge how vital a brand is by its presence in social media. This is why one of Millward Brown ’ s latest innovation projects has been to develop a methodology for measuring the online social vitality of brands.
To monitor the social vitality of a brand we use Twitter. In 2011 45% of Mexican Internet users claimed to be on Twitter (up from 39% the previous year). Twitter is a very useful tool for marketers and market researchers because it acts as a portal into all sorts of different content shared online. And it is an open platform which means we can analyze that content as much as we like.
In order to monitor a brand ’ s online social vitality we have created the VerveIndex: a one-number social impact score. To do so, we calculate Verve for everyone who talks about the brand or its competitors in a certain way, and we weight the scores for sharing and influence. Unlike many other monitoring systems we have leveraged our understanding of brands to create an in-depth measure of social vitality and have dealt with the substantial biases that need to be taken into account with this type of listening.
The end result is a metric that allows marketers to understand just how vital their brand is when it comes to the online domain. Here we see the vScore tracked over time. As with search and viral advertising the level of social commentary is very dependent on other advertising and events. As TV advertising drops away so too does commentary from fans. A FourSquare special boosts comments but fails to sustain that lift. Moreover, in a competitive context, Brand X ’ s lower frequency of mentions counterbalanced out the impact of its positive sentiment in its vScore. The Foursquare promotion was double-edged in that it promoted more mentions and check-ins, but many of those who tweeted about it were not discussing the brand beyond that. In future the brand needs to implement mechanisms to sustain the discussion beyond the initial tweet. Source: Millward Brown ’ s Emerging Media Lab
That was a whistle stop tour of the last 16 years of online advertising as seen by a market researcher. For me, the scary thing is the degree to which a very simplistic metric had dominated our approach to online advertising. If we stick to using click through as the primary metric by which to evaluate advertising response I can ’ t help wondering if there is a threshold at which marketers suddenly come to believe that their investment in online is wasted.
And yet the silly thing is that there is 16 years of evidence that online can have just as much brand building impact on a per person basis as any traditional medium. Whether it is a static display ad, online video or a Facebook page the right research methodology can tease out its brand building effects. And after all, that is what people really buy – brands – not advertising.
To end on a positive note, I am glad to be able to tell you that some of the largest and smartest online advertisers are now beginning to take a more integrated approach to evaluating and optimizing their online advertising. Instead of relying on one metric or another we work with the client and their media agency to help optimize their campaigns based on both behavioral and attitudinal data. The end result is a more effective and efficient approach to managing online marketing spend. Hopefully developments like this will mean that in future online marketing will no longer be hostage to a single behavioral metric and assume its rightful role as a brand building medium.
Thank you for listening.
Presentación de Nigel Hollis en IAB Conecta 2012
Beyond the click: 16 years on Using online advertising to build brands1
The curse of the click Is online at risk of under-estimating its value? 2
16 The number of years since we first proved The number of years since we first proved that online ads could build brands that online ads could build brands4
64% The proportion of advertising and agency The proportion of advertising and agency executives that use CTR to evaluate performance executives that use CTR to evaluate performance5
0.09% The average click through rate The average click through rate reported by DoubleClick in 2010 reported by DoubleClick in 20106
0.08% The average click through rate for aa The average click through rate for BLANK AD BANNER BLANK AD BANNER7
1.3% The average uplift in purchase intent measured in The average uplift in purchase intent measured in response to online ad exposure over the last three years response to online ad exposure over the last three years8
0.02% The correlation between click through The correlation between click through and purchase intent and purchase intent9
39% The proportion of online ads that had aanegative The proportion of online ads that had negative effect on purchase intent in the last three years effect on purchase intent in the last three years10
What is a click? An impulse expression of interest in a brand or offer (often primed by prior exposure in other media)11
All media are capable of doing all things on a per person basis. John Svendsen Global Brand Director Media, Millward Brown12
Brand Building 16 Years On What has stayed the same?13
The three golden rules 1. Make sure people recognize which brand is being advertised 2. Make sure each frame of an ad will work on its own 3. If your goal is to persuade don’t interrupt or annoy18
Ad 1 follows the golden rules to evoke a better response A B C Gusto 14% BC 11% C 7% Relevancia 46% C 45% C 43% Credibilidad 68% BC 61% 65% B Nueva 59% C 55% C 54% información WoM 41% C 39% C 33% (potencial) Efecto viral 47% 60% AC 40% (potencial)19 Diferencia entre creativos A=1, B=2, C=3.
80 60 13% Ads 40 20 0 0 5,000 25,000 20,000 15,000 10,000 The proportion of online ads that achieve more Views per week The proportion of online ads that achieve more than 5,000 views per week on YouTube than 5,000 views per week on YouTube27
We can predict the viral potential of an ad with ABCD28
To get results you need to publicize your content
Online must work in synergy with other media30
Online and Facebook add little unduplicated reach TV Total Online Display Total 84% 35% 26.6% 9.6% 20.3% Online only 1.9% 3.4% 27.7% Facebook Total 4.9% 56% None 5.5%
Each medium had a positive ROI Individual media channels sales contribution to short term sales and media investment 150% Indexed to Investment/Return Ratio: TV Spend 1:1.11 100% Investment/Return Ratio: 1:1.02 50% Investment/Return Ratio: 1:1.34 0 % 111% 100% 46% 45% 19% 14% Sales Spend Sales Spend Sales Spend TV Online Facebook Display Display32
And online adds an important contribution to impact33
What is a like? A superficial indication of brand allegiance (primed by experience of the brand elsewhere)35
Fans are valuable to you…keep them that way 13% 3% Average share of wallet36
FanIndex case study Mexico % increase in opinion Norm Offers something different 47% Better quality than other brands 43% Irresistible taste - For when I have fun with my friends - Worth the money 30% Is the most popular 42% Is a brand I love 46% Setting the trends 45%
Making your fan page flourish Variety Community DIFFERENTIATOR DIFFERENTIATOR Useful Interaction information DIFFERENTIATOR DIFFERENTIATOR38
THE WORLD’S LARGEST It’s open, It’s mobile and it’s an INFORMATION aggregation of all other social platforms PLATFORM: Local/Events Groupon Location Original Tweets Living Social Foursquare EventBrite TripIt MeetUp Yelp PhotosVideos InstagramYouTube yFrogMegaVideo LockerzUstream Twitpic Blogs
VerveIndex: a one-number social impact score FREQUENCY SENTIMENT How often do people talk Is the conversation positive, about you? negative, or neutral? vSCORE SOLICITATION LOCATION What or who prompted the Are people broadcasting mention? their offline actions? INFLUENCE Are the people talking credible on the topic?
42 TV AIRTIME SLOWED DRASTICALLY PROMOTED TWEET WITH FOURSQUARE SPECIAL BRAND XWhat events and topics drove Brand X’s vScore?