Display Advertising Works, But It Works Differently Than Search

There’s been a lot of talk recently that search and other forms of "performance" or "direct" online marketing will continue to grow in the downturn while display and other forms of "brand" marketing will struggle. I think that’s likely to be true at some level but it doesn’t mean that display advertising doesn’t work.

In fact, a recent comScore white paper that was presented last week at the Empirical Generalizations In Advertising Conference at the Wharton School is very enlightening. You can get a copy of the white paper by asking for it here.

comScore operates a panel of over 2mm internet users worldwide and they examined 139 banner ad campaigns and compared a test group that was exposed the the campaign versus a control group that were not exposed to the campaign but are similar in other important ways to the test group.

Specifically, they found the following:

It’s clear that display advertising, despite a lack of clicks, can have a significant positive impact on:
– Visitation to the advertiser’s Web site (lift of at least 46% over a four week period)
– The likelihood of consumers conducting a search query using the advertiser’s branded terms  (a lift of at least 38% over a four week period)
– Consumers’ likelihood of buying the advertised brand online (an average 27% lift in online sales) 
– Consumers’ likelihood of buying at the advertiser’s retail store (an average lift of 17%)

The basic insight from the report is that display advertising does not normally result in an immediate click. That makes sense because the ad is not being presented in a moment of purchase intent, like a search ad is. But the ad does create interest in the product or service which is realized at some later date in the form of a site visit, a search query, and possibly on online or offline purchase.

Here are some charts from the report which illustrate the above findings:

Advertiser_site_reach

The chart above shows that a display campaign significantly increases site visits for up to four weeks after the ad has been viewed.

Search

And this chart above shows that search queries on the advertiser’s brand or trademark increase significantly for up to four weeks after the display ad is viewed.

One of the most interesting findings is the following:

The impact of search ads alone on consumers’ buying behavior was found to be clearly greater than that of display ads alone. This is true both in terms of the ads’ impact on online buying as well as the impact on offline sales. This is not surprising because consumers responding to search ads are much more likely to be “in the market” for buying the advertised product. However, it must be remembered that the reach of display ads is typically much higher than that of search ads. For example, in the studies conducted by comScore, approximately 81% of the consumers who saw an ad received only a display ad, while a much lower 8% received only a search ad. When the lift factors are weighted by the reach of the ad, display ads typically emerge as being able to generate a higher total lift in sales.   

Conventional wisdom says that media works best when multiple programs are used together, and we do indeed see clear synergies between search and display. Both in terms of the impact on buyer penetration and dollar sales (per thousand consumers exposed), it’s clear that the combination of search and display together is greater than the sum of the impact of display and search ads separately. The synergy is notable.

Search_and_display

So this means that the best approach is a mix of display and search. That is what most marketers are doing and they should continue to do it even in the downturn. However, this study does not determine what the right price for display advertising is. Search is measurable by virtue of the cost per click and display is not. And if click thru isn’t the right measure, then display has an issue. Not every campaign can be compared in a controlled test as comScore did in this study. But comScore could make their panel available to advertisers to measure performance of display campaigns. I am no longer involved with the company (I was on the board for almost 10 years), but I expect that they’ll do something like that. And it will be a great help to the display market. Particularly in tough times when every ad dollar spent needs to count.

Reblog this post [with Zemanta]
#VC & Technology

Comments (Archived):

  1. curiouslypersistent

    Very interesting, thanks for providing the link to the white paper.The findings make sense to me. Online advertising is often presumed to be just about clickthroughs and CPAs. However, it can also be used as a branding mechanism in conjunction with other media.Dynamic Logic have done some work – this being one example – that shows that the biggest effects for online advertising are in awareness and imagery, not in purchasing.The internet is a fabulous mechanism for measuring the impact of sales for DR campaigns, but this is far from a one trick pony and the other elements shouldn’t be forgotten(Of course, brand impact is far more difficult to measure, and that gets into a whole new can of worms)

  2. GlennKelman

    Fred, I’ve argued before that display ads haven’t been as effective at bolstering a brand because they don’t have the emotional impact of a glossy magazine or of a TV commercial. This is probably partly because back in 2006, Redfin could never sort out how to make display ads work, and partly because we don’t run ads either.But notwithstanding that display ads generally are getting prettier and better, especially when they run as video, the fact is you have convinced me that I was wrong. Your posts on this topic have really been fantastic.

  3. Matt Meeker

    Very interesting. It seems the element of frequency is missing here though (I admit, I haven’t read the study yet, just your post).Wouldn’t the results vary significantly if I saw a display ad 100 times in a one week period versus seeing it only a single time? There are definitely those ads that just “stick” with you, but mostly because you see them over and over again in a small space of time.Since you’ve read the study, does frequency factor in here, and what is the point where an ad has enough exposure to become memorable?

    1. fredwilson

      there is more work to do here and the study says:”There are additional factors that may affect the rates of lift over time that are not addressed in this paper. The most obvious include the length and frequency levels of each campaign. The average (mean) length of the campaigns included in this study is 42 days. The shortest campaigns ran over a single day, and the longest campaign ran for 108 days.Another consideration is the purpose and type of the creative message. “Call to action” or direct response ads that are typical of sale-related retail campaigns clearly generate a different level of response than “branding” ads designed to elevate awareness and build a brand.”

      1. Matt Meeker

        Really good learning there. Thanks for sharing!

  4. michaeldowns

    CPA (and some CPM) ad networks place cookies and negotiate a hold-over period to ensure that they get credit for sales generated up to 30 days after the initial impression(s). While each network tracks it’s own follow-on purchase performance, it’s rare to see such comprehensive information in public. Having said that, these numbers inflate the lift that I’ve seen by 25-50% raising questions about how the study was conducted. One variable that has a pronounced effect on follow-on purchases is the quality of the display inventory. Not surprisingly, there’s a big difference between Yahoo’s main page and Facebook’s ROS.

    1. gian Fulgoni

      In measuring the impact of display ads, we parsed the comScore panelists into two groups: 1) those exposed to at least one display ad impression (the “test” group) and 2) a “control” group not exposed to any ads. The control group panelists were selected to be similar to the test group in behavior (prior to the start of the ad campaign) and demographics. We then computed the lift by observing and comparing the behavior of the two groups following the execution of the ad campaign through the use of comScore tracking software that’s installed on each panelist’s machine. We do not use cookies because of the problems created by cooke deletion.. Four independent studies (one of them conducted by comScore) have now shown that at least 30% of Internet users delete their cookies in a month. So, if the lift numbers you cite from the ad network studies were based on tracking cookied machines, they will undoubtedly have understated advertising’s lift by a factor of at least 30% because they would not have been able to take into account the machines whose cookies were deleted

    2. Tim Daloisio

      Yes….but there are significant issues with pure cookie based tracking in regards to display ads.1. Cookie deletion rates are aggressive, 10% or greater flat out don’t allow them, 35% plus are deleting on a weekly basis.2. Cookies only work when all activity is on a single PC. How many of us use multiple devices to access the internet for purchase information.

      1. Paul Cook

        Just thought I share a study we did with one of the worlds largest advertisers using FSOs to track campaigns and comparing to cookie figures. As a lot of advertising is now being sold with cookie based targetting it follows that everybody who sees the ads will accept cookies to some degree. Some only accept 3rd party cookie ( AKA Safari users as this is the default settings) but otherwise cookie deletion was a actually fairly igsignificant from a statistical point of view for low frequency campaigns. That said if you have a high frequency campaign with certain audiences we found some days where half the people the campaign reached had deleted their cookie. Of course all this doesn’t get round point 2, I believe the average is 2-3 machines per users and know banks that have measured it at around this level for their online Banking.

  5. fredwilson

    All good questions and I think this methodology could tell us a lot morethan it has to date

  6. CardiffGiant

    With stock returns, the single biggest determining factor in your long-term return is the price you paid for the stocks. If you invested in 1999, your long-term return is going to be fairly ugly. If you invested in 2002, you’ve got a much better chance of achieving above-average returns.And so it is with advertising returns. This white paper, while interesting, doesn’t address the issue of the cost of display advertising so it’s impossible to have a conversation about its effectiveness. You need to know the denominator to calculate ROI. So while, on first blush, a 17% lift in likelihood to purchase from an advertiser’s online store is positive, it’s not positive if the cost to acquire that 17% lift exceeds the economics of the benefit of that 17%.I’ve run mult-million dollar campaigns with both search and display ads and I’ve never seen display advertising generate a measurable positive ROI. That’s not to say it doesn’t have a place, but I don’t think you can make the case it’s at all like paid search when it comes to quantitative results.

    1. fredwilson

      agreed, that is why i wrote this near the end of the post:”However, this study does not determine what the right price for display advertising is”i suspect there is a right price and my hope is panel based measurement can help determine that

      1. val_popov

        Clearly, a new metric is needed. One that measures actual ad viewership, which is people who pay attention to the ad. The question is how to distinguish between people who ignore the ad and those who pay attention. I expect we will see quite a bit of innovation that tackles the issue, especially as the shift away from CPM continues to put pressure on display advertising. Ultimately, such a metric will allow pricing per actual ad view. If such a metric is delivered through the context of a social network or a user community, it could be coupled with user identity and demographic data, which could enable the advertiser to lock in the cost of a specific exposure to a target customer.

    2. gian Fulgoni

      I agree that it’s important to know the cost of the display ad campaign and to measure its ROI. That said, smart marketers are realizing that if they can get the publisher to accept a “pay-for-performance” deal that’s based on the number of clicks, then they are virtually assured of getting a very attractive ROI. The reason is because average click through rates on display ads have dropped to under 0.1%. So, with the kind of lifts we’re seeing and minimal clicks the advertiser gets a terrific deal. But, the publisher gets screwed.

      1. VJ23

        Could you or anyone help me interpret the last chart — specifically what does “online $$ per 000 exposed” mean? Does it mean that, say spending $20 per CPM on average results in = $1,263 – $994 = $269 in extra revenues?

  7. gmarch

    of course control vs exposure shows lift in display ads. They cost money, they should do SOMETHING. The “see we’re not useless” numbers don’t move me as a marketer.Do display ads “work” more than other options like TV, Radio, OOH. When do they work harder? Maybe its different according to target audience, or the product category, or the timing of delivery.A more elegant question would be at what frequency do these other mediums show diminishing returns in the metrics you care about, such that adding another impact focused tactic (like display) is additive. For example does the 8th TV exposure generate more or less lift in metrics than the 1st display ad? Or vice versa.Does it work is a clunky question when it comes to media choices. Standing in Time Square with flyers works to sell discount suits, but should Snickers do that? Is that the best utilization of all of their options?

  8. johndodds

    The use of a 90% confidence level fills me with little confidence.

  9. Ryan Poll

    I’m on vacation reading this on my iPhone so I’m sorry if I missed it, but did the paper talk about targeted ads vs non targeted?

  10. fredwilson

    Great comment DarrenThanks for stopping by and sharing your thoughts

  11. jasonkolb

    This is great news for online advertising: it does actually have a purpose. However, current company valuations are based on advertising that actually causes sales, so that means you can drop the value of any advertising-based company by about 50% right there.

  12. Darren Herman

    Thanks for showcasing this data from comScore. Randall Rothenberg, CEO of the IAB also wrote recently about this (and some tangentially fascinating information): http://www.randallrothenber…The big issue we have with online advertising is that we’ve been looking at it from a performance point of view for so long (in the 90s, it was educating about the clicks) but now, we realize that view-thru (not click-thru) accounts for quite a bit and the attribution of how we divvy up online advertising is still in nascent stages.Please note that most major media agencies and brands use different methodologies on attributing performance. This is generally done at the client level and can range from very short attribution windows to very long. Clicks and views are counted together or can be counted separate, and when counted separate, it’s fascinating to see view-thru attribution across a yield curve.Because people recognize that display advertising works (which I’d certainly argue YES), we are seeing the titans of MSFT and Google work on Engagement Mapping. The Atlas Institute has done a lot of research in this area: http://www.atlassolutions.c…One of the main bottlenecks we face with online advertising is that it should be part of the advertising conversation: not separated to online vs. offline. The genius and strength behind advertising is when you can measure across all marketing channels and not strategize/plan/buy separately but rather, all together to make all marketing channels work in concert together. That is where the future of all digital media is going from an agency point of view. Just having a digital group is not enough. Integration is key and where the bigger ideas will be born from.

  13. kenberger

    That’s just it: display campaigns are tough to measure for results, particularly because their impact tends to happen over an indeterminable, very variable time. Yet according to a McCann Erickson exec I just met with, brands that think they can forego the exposure they get in this way do so at their mid- to long-term peril.The effect is even greater in places like Japan. And for example, Nikon USA recently got a Japanese-born CEO. You can immediately see the effect via big increases in their media campaigns to boost the brand, including increased celebrity appearances, etc.

    1. vonsydow

      Great post Fred, it’s always refreshing when non-ad industry people take a stand on behalf of the long term effects of ad exposures ;)Back in 50s and early 60s the pinnacle of great advertising was considered to be historically proven methods, pre-testing and analysis. Create, measure, evaluate and adjust. Rinse and repeat. Sound familiar?This was, naturally, before advertising had it’s “creative revolution”, when Bill Bernbach and others showed the world the power in allowing ad creatives to seize power from statisticians and MBA’s, showing that intuition and imagination, not pre-testing and proven methods, could create quant leaps in great advertising.Over forty years has passed and once again, we’re living in a metric driven, cost-per-click world, where the Internet, despite all it’s creative promise, is increasingly becoming a channel primary for direct-response marketing. “Demand fulfillment”, not “demand creation”. Sure, the Internet is the most precise marketing vehicle ever created, but can you even when was the last time you saw a really great web ad?The Internet has yet to experience it’s creative revolution, which is partly due to ad agencies’ lack of tech knowledge. When we presented at TC50, Marc Andreesen pointed out that ad agencies are “notoriously technology averse”. I live in Sweden, so I’m not sure if this reflects the feeling of the majority of the technology community “over there”, but I get the sense that it might.However, I think it’s unreasonable for us tech people to expect agencies to keep up with the pace of change for all channels they utilize. They’re focused on communication, not technology, and as long as we expect them to work with tools and interfaces designed for completely different skill sets and lead times (apps and sites), my guess is that we’ll continue to see print and TV ideas adaptations, complemented by a “click now” button ;)So. It’s up to the people that read blogs like this to create products that decrease the gap between creativity and technology, so that ad creatives can make full use of what our medium has to offer. There are many companies that have promising ideas for improved advertising, specially those working on semantic tech and real-time “data smart” ads. But having looked how they’re packaging their products it’s clear to me that they, like most tech companies with a very rational developer view of the world, have very little knowledge of the workflow in creative advertising agencies.The NYT said it very well the other day:”There’s no doubt that there will be a lot of data that can be collected that could be applied to the creative process [but] that’s not necessarily an easy discussion to have with great art directors.”Coders and copywriters think very differently. And unless we acknowledge this, and start creating products that reflect how “demand creating” advertising works, we’re destined have the search vs. display debate for another two years, continuing to miss the opportunity to trigger version 2.0 of the creative revolution, so that digital media can see the same explosion of advertising ideas we did when TV saw the light of day in the 60s.

  14. Wallen's

    Extremely interesting Fred. Thanks.It would be interesting to put these figures in perspective with the standalone impact of TV/ radio/ newspaper ads. The real power of internet ads lies when all channels are coordinated with traditional media campaigns. I had seen some research from McKinsey (sorry no link) showing that web ads had more impact early in the sales channel, e.g., during the information gathering phase.

  15. Randall

    I’m not really sure what the mystery is here. Commercials on television enable NO click, yet plenty of research and annecdotal evidence indicates that advertising pays. Coca cola is the case study.Still though, Internet advertising is lame. If you have a good product you don’t need to advertise. Again, coca cola is the case study here as well. It’s a crap product, though sells plenty in spite of the fact that water is much better for you and costs much less.Why do people make these irrational decisions?Name recognition. I did a study in college. No one knew me. I wanted to know if I could win a college political campaign through name recognition alone. I plastered the target audience with ads. Flyers mostly. Sure enough, I won — even though no one had any rational reason to vote for me.Old hat. Easy game.Changing the world though innovation and a genuine interest in making it a better place is much harder and also much more rewarding.Imagine how coca cola employees would feel if they had the #1 soft drink without advertising.Different ball game isn’t it?What’s the case study here? The Toyota Prius.Toyota can’t make enough of the cars. Dealers were, not sure if they still are, charging well over sticker with a huge deposit just to get one. This in spite of the fact that the majority of advertising for automobiles is from the SUV portion of the market. Big trucks and high margin vehicles dominate the commerical airwaves, yet Toyota can’t make enough of the vehicles they advertise relatively little. Actually, if you watch television, you will probably never see an ad for a Prius and still they sell like hot cakes.The advertising / marketing game is an easy one. Build a product, sell ads that promote its weaknesses as strengths and conjure an image in the mind of the person that is the opposite of the product itself. When the viewer of the ad encounters the actual product, they are left to compare the image of the product in their mind with the actual product itself.Now we get into human nature. Humans don’t want to be duped. They’d rather continue to believe coke is a great soft drink or an SUV is a good vehicle that change their mind to one of true understanding. There is no rational choice that is not the Toyota Prius, yet other vehicles sell. Why? Name recognition.Advertising sells. It’s an easy game. It’s a known game with a researched, studied, and documented set of rules.On the internet the rules change — they almost disappear completely — but name recognition still sells. People feel more comfortable with what they know than what they don’t know. If someone has never heard your name, they are less likely to engage in conversation than if they have heard your name 1,000 times.I’m not really sure what the mystery is here. Commercials on television enable NO click, yet plenty of research and annecdotal evidence indicates that advertising pays. Coca cola is the case study.Still though, Internet advertising is lame. If you have a good product you don’t need to advertise. Again, coca cola is the case study here as well. It’s a crap product, though sells plenty in spite of the fact that water is much better for you and costs much less.Why do people make these irrational decisions?Name recognition. I did a study in college. No one knew me. I wanted to know if I could win a college political campaign through name recognition alone. I plastered the target audience with ads. Flyers mostly. Sure enough, I won — even though no one had any rational reason to vote for me.Old hat. Easy game.Changing the world though innovation and a genuine interest in making it a better place is much harder and also much more rewarding.Imagine how coca cola employees would feel if they had the #1 soft drink without advertising.Different ball game isn’t it?What’s the case study here? The Toyota Prius.Toyota can’t make enough of the cars. Dealers were, not sure if they still are, charging well over sticker with a huge deposit just to get one. This in spite of the fact that the majority of advertising for automobiles is from the SUV portion of the market. Big trucks and high margin vehicles dominate the commerical airwaves, yet Toyota can’t make enough of the vehicles they advertise relatively little. Actually, if you watch television, you will probably never see an ad for a Prius and still they sell like hot cakes.The advertising / marketing game is an easy one. Build a product, sell ads that promote its weaknesses as strengths and conjure an image in the mind of the person that is the opposite of the product itself. When the viewer of the ad encounters the actual product, they are left to compare the image of the product in their mind with the actual product itself.Now we get into human nature. Humans don’t want to be duped. They’d rather continue to believe coke is a great soft drink or an SUV is a good vehicle that change their mind to one of true understanding. There is no rational choice that is not the Toyota Prius, yet other vehicles sell. Why? Name recognition.Advertising sells. It’s an easy game. It’s a known game with a researched, studied, and documented set of rules.On the internet the rules change — they almost disappear completely — but name recognition still sells. People feel more comfortable with what they know than what they don’t know. If someone has never heard your name, they are less likely to engage in conversation than if they have heard your name 1,000 times.”Oh you’re from Google??””Oh you’re from … how do you pronounce that again?””Oh you’re from Google??””Oh you’re from … how do you pronounce that again?”99% of the world doesn’t even know Google is misspelled and they don’t care because they know google. This is the power of name recognition. Google got there not through advertising, but through a good product. A good set of search results.On the internet, the rules change. Sure, you can sell a product through advertising. Anyone can sell a product through advertising, but why would you want to when it’s much more entertaining to build a good product?

  16. TedHoward

    I’ve seen a few other instances recently of numbers and commentary along these lines. The first such news I saw was Microsoft/Atlas’s Engagement Mapping PR. It’s good so see more research coming out. Due to the economy, there may be a drought of innovation, but it seems to me that different types of advertising should be always be treated differently until proven otherwise. Social network advertising, display, search, mobile, email spam – these all have their own quirks, inefficiencies, etc. I suspect that in any large organization, each one will become its own sub-specialized profession eventually.http://www.atlassolutions.c…Extra juice for my ‘all ad types are sub-specializations’ thought:I just read a bit on Alley Insider about P&G’s experiences with marketing on Facebook. The AI’s or Nicholas Carlson’s suggestion is for Facebook to just do sponsored takeovers rather than trying to find new marketing mechanisms. I would suggest both, maybe bundling such that the sponsorship deals are contingent on also paying Facebook to try crazy new ideas. My two cents anyway.http://www.alleyinsider.com

  17. Jim Peterson

    Fred, thanks for pointing to the ComScore report. I had written off display advertising but will read this report and take another look. Thank You.

  18. Justin Seibert

    @Fred – nice post. You make several good points, particularly about how well search and display can work in conjunction. What will be very interesting is how aspects of both can be managed under one account as the paid search accounts (Google in particular, with Yahoo! coming on, too) allow more refined use of their content networks.

  19. Brad Cooper

    Thanks, Fred. Glad to see this new data come out as it’s been a couple years since I’d seen similar analysis done by Atlas/AvenueA/Microsoft, Nielsen, etc.The points above on “but, how much did that lift cost?” are valid. Not only can’t we just say “display doesn’t work b/c I can’t measure it” we also can’t say “display works, I just can’t measure it.”But, I’ve can add my own personal experience to the mix here… I’ve seen firsthand the combined impact of display and search in some tests my team did at one my prior companies during several large campaigns. Not only was the impact of display on search sizable, but we determined that it would be next to impossible to reach our customer acquisition goals without it. Search is fantastic, but it has a pretty hard ceiling. You can optimize like crazy and expand keyword lists, but search gets to a point where it just doesn’t grow fast enough to fully replace traditional media. Display allows for more scalable volume and has the dual impact of boosting search and other media types, including traditional.

  20. Paolo

    Mmh, I’ve been working in the industry for over 10 years now and I could mention tons of similar “studies”. I won’t consider the most obvious objections, that BTW have already been made (frequency, duration, pressure, cost and so on). Another very important factor would be the presence of those advertisers on other media (TV, print, radio) while the “experiment” was conducted, but again, these are just details.IMHO, these kind of studies are simply useless and misleading, ’cause they’re made to sell more display ads, especially in though periods, such as the present one. Truth is that the Web is a direct marketing medium. TV, radio & press are much more powerful and cheaper (in terms of cost per contact) to run a branding campaign.So, do you want to sell more ads? That’s fine for me, but please don’t ask me to believe that “display campaign significantly increases site visits for up to four weeks after the ad has been viewed”. I can still distinguish between a “study” and a “sales pitch”…

    1. gian Fulgoni

      Paolo:The study was funded by comScore and we’re not in the business of selling display ads, so I can put your concerns about bias to rest. I also want to repeat my earlier comments that you might have been missed:In measuring the impact of display ads, we parsed the comScore panelists into two groups: 1) those exposed to at least one display ad impression (the “test” group) and 2) a “control” group not exposed to any ads. The control group panelists were selected to be similar to the test group in behavior (prior to the start of the ad campaign) and demographics. We then computed the lift by observing and comparing the behavior of the two groups following the execution of the ad campaign through the use of comScore tracking software that’s installed on each panelist’s machine. We do not use cookies because of the problems created by cooke deletion.. Four independent studies (one of them conducted by comScore) have now shown that at least 30% of Internet users delete their cookies in a month. So, if the lift numbers you (Michaeldowns) cite from the ad network studies were based on tracking cookied machines, they will undoubtedly have understated advertising’s lift by a factor of at least 30% because they would not have been able to take into account the machines whose cookies were deletedI agree that it’s important to know the cost of the display ad campaign and to measure its ROI. That said, smart marketers are realizing that if they can get the publisher to accept a “pay-for-performance” deal that’s based on the number of clicks, then they are virtually assured of getting a very attractive ROI. The reason is because average click through rates on display ads have dropped to under 0.1%. So, with the kind of lifts we’re seeing and minimal clicks the advertiser gets a terrific deal. But, the publisher gets screwed.

  21. Paul

    I think the results of this study are interesting and I think comScore has done a good job of initiating a discussion on the topic (along with your post Fred). The only methodological issues I have with it are that:a) I think you need to make a distinction between advertising for a specific product and advertising for an online e-commerce sites that sell multiple products/services. Display ads for Best Buy remind me to go to Best Buy’s site over the holidays to buy gifts – it’s a lower bar for action as I might have multiple products I am interested in on their site compared to a single product/service advertised on another site.b) The control group displays a healthy growth rate (in visitation to advertiser’s site) over time. This strongly indicates that there are other components of the campaign that they are exposed to (offline media?). This also indicates that the test group may have been exposed in a similar way making the extra lift from display ads actually extra lift from display + <something else=”” we=”” didn’t=”” can’t=”” control=”” for=””>. This doesn’t invalidate the findings, it just means there may be other media synergies playing into the strength of the display ads. For instance it stands to reason I am more likely to go to Best Buy’s site if I see both a billboard and an online display ad from them – especially if both exhibit similar creative.c) It worries me a little that the results presented are a ‘study of studies’. comScore has done a good job of breaking down the categories to show you the number of studies in each category and the average ‘lift’. However, I’d love to know a little bit more about the variance of results by study in each category. Just to be confident we aren’t seeing wild swings and that all studies do indeed show positive, not negative results. A negative result would put a different spin on the interpretation – that there are circumstances in which a display ad produces no lift whatsoever, why?Overall though, from the studies I have seen of this sort, it’s a great bit of research. It’s as detailed as you need it and they have definitely crossed their ‘i’s and dotted their ‘t’s regarding design. It wouldn’t be hard to extend this design to include a cost analysis.

  22. Peter Daboll

    A couple of quick comments. We have found several similar results, that there is a strong, often unmeasured impact of display advertising on brand and sales metrics that are almost inversely related to CTR. in other words, CTR is a terrible metric for the efficacy of a brand ad.Second, I think beyond looking at visitation or reach as the important variable of success for a display ad, i believe it is even more important to understand, measure and drive the consumer’s engagement on the site. As many Comscore analysis have taught us, most sites visitors are “accidental tourists”. Visitors with low levels of engagement with the site who often visit the site just once. Being able to measure and increase customer engagement has measurable, long-term positive brand effects that result not only in increased sales of the brand for that household, but an amplified sales effect across that users’ scope of influence. Engagement is ultimately where the value is for the brand.

    1. Randall

      What is a good metric for the efficacy of a brand ad?

      1. Paul

        That it made consumers/customers/viewers/listeners think and feel about your brand the way you intended . That’s a ‘first level’ effect. You can go further and try and link interest, behavior, engagement ‘with something’ as further effects, but at the end of the day, if your brand ad didn’t achieve its communications goal, it probably isn’t working.

    2. fredwilson

      Hi Peter ­ nice to hear from you. I hope you are doing well!

  23. Matt

    Hi Fred,Couldn’t agree more! Your post fits in perfectly with my post on the official Yahoo! Search Marketing blog:http://www.ysmblog.com/blog

  24. fredwilson

    Some great ideas in this comment

  25. GavinH

    Hi,Perhaps of relevance is Display re-targeting (behavioral) also called remarketing used in conjuction with pixel/tags tracing individual potential buyes. On Blue Lithiums network- the lines in a digital campaign canlook for new traffic/unique users to a web site + specifically look for people who ‘nearly’ converted, see link belowROI can be measured in Display – just needs a different perspective..they do have to generally start out on a CPA basis with Yahoo but you can definatley work out a CPC after a few days in an optimization/intial phaze…CTR is also another point of difference – (.10 would be an average CTR for a Travel vertical campaign) – the CTR of display may pale into insignificance in a direct ‘apples to apples’ comparison to search CTR’s (with an average upwards of .75%+). If a campaign places a value on its CPA – and – if it is using pixels to track conversions/CPA effectively, providing there is a good call to action on the creative, you are likely to drive down the cost of your CPA using remarketing mentioned above…research shows that it is Far easier to get a consumer to eventually convert (by retargeting them- say a number of times within 30 days) as opposed to going away + bringing greenfield traffic from scratch……..see “Ad Parth Remarketing” piece here on Blue Lithium’s website….http://www.bluelithium.co.uk/core_… , hard to get specific evidence from these guys as they are IASH compliant to keep their network blind.Still though- was reading this article + these comments + I thought I would throw it out there….

    1. fredwilson

      Thanks for this. Retargeting is a great way to use banners effectively. Thanks for pointing that out

  26. Samuel Lavoie

    Great article, was late on that since its 3 months old but still, great finding. And more importantly, it open the discussion about how we price banner/display advertising and how the current model could be broken.I will definitely see a strategy like going for banner advertising in your niche starting with a big chunk of budget and going toward a more SEO-PPC tactics over the months expecting a 80-20 shift over a 6 month period.Great post as a reminder that display advertising is still and need to be an important part of the total eMarketing package.

  27. Brad Krassner

    The problem that exists today with display ads is the publisher can not tell you how many times your ad has been viewed. They can only tell you that a Log File has recorded so many ad requests that were loaded onto a web page. This is what is causing the confusion and mistrust among advertisers. The Media Rating Council is finishing up an audit to certify a new technology call RealVu that can now track all ads as the are requested as the are rendered onto a page as they appear in the viewable area of the browser screen and for how long the Ad appears in the viewable area. You can see this technology for yourselves at http://www.realvu.net This new technology is going to change everything for display advertising because advertisers will only pay when their ad is viewed not requested. Knowing if an ad was viewed and for how long will also make reach and frequency data reliable and trustworthy and hence will bring more dollars into the display ad market.

  28. edwardizzys

    Interesting!In the past, offline media was one of the major mediums to promote one’s business and brand awareness. However, with the increasing popularity of the online media and its positive results, more people are turning to online marketing techniques to build a strong business presence across the globe. There are several online marketing strategies we can use for our Internet marketing solutions, but SEO (search engine optimization) today has taken a stance much ahead of any other technique.SEO brings with it tremendous scope for business improvement with improved keyword positioning, improved ranking and overall increase in the ROI and cuts down on the PPC cost extensively in the long run. This is a proven marketing solution if done the right way using the right techniques based on your website requirements.I thought this information would be helpful for those who wish to promote their website and get great results.