Jonathan has helped optimize...

« Links vs. Cookies: A Tale of Two Web Economies | Main | Yieldbot is Born to Run »

Display’s Matching Problem

Puzzle

Simply put, advertising is a matching problem.

It is a puzzle of people and needs, demand and awareness, creativity and messaging, and most of importantly, timing and context. It works best when the shape of the consumer fits perfectly with the shape of the advertiser. While conceptually this idea is not new  - advertisers have long been armed with reams of information on all these data points - matching in the legacy of advertising was at best a fishing excursion and at worst a waste of buckets of cash.

The web changed everything. All of a sudden we had the other piece of the puzzle. As the first medium that was actually controlled by the people consuming it, that very control became available to advertisers to match against. People were raising their hands and announcing what they were interested in doing and discovering. This of course made the matching pretty easy. Search grew the tool sets to take advantage of this. Amazingly, publishers with their loads of site analytic data did not.

The inevitable result when all media is accountable and the puzzle pieces fit together is that all the wasted dollars dwindle and the dollars not being wasted end up on a matching platform. And if the platform is free, real disruption occurs.

By 2003 with Craigslist available in 32 US cities publishers were feeling the first effects of easier matching. Classifieds are the original matching advertisements so in hindsight it was logical that they would be the first revenue to get disintermediated by better digital matching systems. In this era of realtime it sometimes gets forgotten that not only the immediacy of digital publishing caused old media’s demise. More than anything else the ability to efficiently facilitate consumer-to-advertiser matching killed old media.

Fast-forward and now it is display advertising that is old media.

The first thing to remember is that display was born as old media/old advertising’s solution. The idea was started by a print magazine and embraced as an easy way to lure ad dollars into an emergent channel. Not much has changed with display since that time. That’s because the way its technology has been built makes it incredibly difficult to match consumers with relevant ads. There are some verticals where appending data may be helpful but as mentioned earlier, timing and context are critical in shaping a match.

This gets to the root of display’s matching problem. The temporal value of data is fleeting and context is frequently not discernable. If we look to Search for matching guidance context is exactly what Google’s algos are trying to understand. It’s why they take in-session query history into account. It’s why they look at temporal factors from time-on-page to time-to-click. It’s why they take landing page bounce rates and copy into account. It’s why they’ve added social factors to SERPs. It’s why 25% of all SERPs (and growing) are being personalized. Context, the weaving together of interrelated conditions that brings meaning to why a person is doing something, is the key to delivering relevance.

Amazingly context is a word I rarely hear anymore in display. In fact I’ve read quite a few people dismiss its value. That’s bullshit. I understand, display was not built to deliver context but I’m afraid I see too few people working on solving this problem – a problem that fundamentally addresses the value of the media. The rise of audience buying is exacerbating this problem. It’s a red herring to get media dollars into the channel because no one is clicking the ads. Why, because they are not relevant. There is poor matching.

The lessons to be learned from the rise of Search extend far beyond the way media is priced, bought and sold. Without new methods and technology to deliver relevance all those buying & bidding platforms and exchanges will be short lived. Sure, there might be some nice exits as there has been in the past, but display ads will still suck, real value will not have been added to the medium and Search will only become more dominant.

This may be the final bell. As with the rise of Craigslist & Google the real losers in this matching game will not be the advertisers. The real losers with display’s matching problem will be publishers. The only thing of true value in a user-controlled medium is relevance. As all media is becoming searchable, time is running out to deliver it.

Previous Related Posts:

Lessons in Digital Content Creation & Technology from Search Advertising

Ads as Content – Content as Ads

TrackBack

TrackBack URL for this entry:
http://www.typepad.com/services/trackback/6a00d8341ccaa353ef01310f59596d970c

Listed below are links to weblogs that reference Display’s Matching Problem:

Comments

Jon, as always good thoughts -I am personally down on Display right now - as you say - it is OLD MEDIA and there really hasn't been any innovation around the banner unit itself - there have been advances on targeting & how you buy them, but the actual composition and make-up of the ads haven't changed radically in the last 15 years...

Keep up the good work!

Jon - Great Post! I don't necessarily agree that there has not been advancements in banner advertising or in the ad unit itself. There has been a lot. BUT... those advancements are definitely minimized by not having the capabilities you mention. The publisher side of the equation needs to get more advanced for sure. Right now the matching advancements seem to becoming for ad networks and exchanges.

Wow - you've written a compelling manifesto for publishers, agencies, and all the players that sit between them in the advertising value chain. I was particularly taken by the emphasis you placed on the importance of timing in shaping a match to deliver relevance. As you say, "the temporal value of data is fleeting". I'm obviously biased in my perspective, but I couldn't agree more, and I believe that it's a logical extension of your thesis to say that the technology that is required to solve this problem must place a strong emphasis on rapidly analyzing large volumes of data, because this is critical to increasing targeting precision (and therefore relevance). In other words, really cool matching technology that takes too long to analyze old data won't cut it. You mention Google as an example of a firm particularly focused on this issue. This paper from a few years ago describes Google's approach to improving the relevance of recommended news articles - your readers that want to get a little geekier on the topic may find it of interest: http://www2007.org/papers/paper570.pdf I thought it was interesting because the results seem to fairly explicitly demonstrate that reducing data latency increases targeting precision and relevance (they produced a 38% increase in CTR versus the benchmark).

Brad - Thank you for the comment and amplifying a critical point of the post. A great link as well. I think it shows how Google (testing these ideas 3 years ago at scale!) is more ahead of the game than people give them credit for. It's great there are others like Netezza that focused on solving this problems for digital media.

Scott - my new saying to display folks is that "if the end result of your technology is a 300x250 display ad then you are not revolutionary."

Steve - you are right, there have been advancements, IMO, mostly on the creative end (Google's purchase of Teracent as evidence). Still, we need even more outside the box thinking to fix this channel, esp for pubs.

Hi Jon, I think you're spot on. While targeting options for search engine PPC has evolved significantly over the last 5-10 years, providing users with a wealth of tools to track, measure and optimise keywords, ads and match types, display has failed to see any significant advances.

Why, in 2010, do Google advertisers still have to throw a whole bunch of similar keywords together in an ad group, then cross their fingers and hope their display ads get matched to the right sites at the right times?

OK, display targeting options have improved recently, but giving advertisers more tools and a higher degree of control to help optimize display ads, such as advanced keywords targeting (i.e. being able to show on sites which mention certain phrases or combinations of words in certain orders or placements, or being able to choose where on a page you show) will surely help improve relevancy.

Display isn't dead, and it can be done well - just look at Facebook. If you're a boat salesman based in Sydney, it's possible to target males aged 35-55 who work in Sydney and enjoy fishing at the weekends.

More inovation from Google in its display network, giving advertisers more control, and allowing them to find their niche (however small), would be a good start to help drive this industry forward.

Hey, Great post!

In this day and age taking advantage of SEO and social network websites as effective marketing tools for your business is essential! Most businesses don’t fully understand the capabilities of these tools to improve traffic and exposure! I was searching for a company to ship one of my vehicles overseas and ran into Auto Shipping Network; I was amazed at how they implemented these tools so well into their company website. Check it out for yourselves at http://www.autoshippingnetwork.com/ .

Jonathan,
I have been one of display's biggest anti-fans for a number of years. I am glad to see someone write what I think a lot of us have been trying to put into words. It seems like display's biggest weakness followed it from the old medium from which it was born.

The only hope I had for display was the idea of "widget" ads--such as searching Kayak prices from a widget in the NY Time's Travel section. That instant gratification/interactivity had me hopeful, but it never seemed to take off. I can see how technologies like Teracent are a step in a better direction though. How to figure out context though??

I like Alan's comment about Facebook, but in my field I still find it difficult to leverage demographic targeting alone to our clients' benefit. My company specializes in real estate advertising. For communities that have apartments to rent, it works *great*. When it's a community of places for sale (i.e., you need to take out a mortgage), we have yet to find a slam-dunk case of it "working". Which, for a web property that (purportedly) has more than 250 million active users, it would you should be able to somehow find a few of those people looking for a new home.

One problem is that there are so many Facebook users who don't really fill out their demographic data, which works against us matching ads to their interests.

Also, adding some behavioral (read:intent) data to the demographic data could help a lot. This is native to search, because as someone put it, search engines are a "database of intention". But for a company that already has so many perceived privacy issues (Facebook Beacon, anyone?) and has to toe such a fine line I find that difficult for them to do.

A number of companies will face similar scrutiny. Google is trying some behavioral targeting, but faces much opposition.

Any other thoughts on achieving better relevance?

Finally, someone who gets the need for "Context". It's critical if you are to solve the matching problem.

4 years ago we started working on a way to deliver real time context to any web server from any device. We started with Mobile. We now have a way to deliver real time context (anything that the user desires to share (with suitable privacy protections) to any web app that can use it.

There are zero behavioral changes required for either the consumer or the web admin. It uses all all approved standards and leverages all existing programming approaches.

Now you can really can match correctly - as long as the consumer trusts you and is prepared to share the data with you.

Cheers,

Peter
5o9 Inc.

Great thoughts. And very well said. Internet marketing has gone far from where it first started. Businesses should take advantage of what is considered as an essential tool and that is the SEO and social networking sites. People should keep up with the fast-paced world of technology.

The comments to this entry are closed.