In Part 1 we looked at the current cash cows. Or, to put it more brutally, the current single cash cow of search advertising that belongs to Google. In this post we look at the current wave of semantic web startups that focus on advertising. As this field today is a bit limited, we also look more speculatively at where the future disruptive technology may come from.

Photo, courtesy AgilityNut on Flickr.

Semantic Is Not The Same As Contextual Or Behavioral

But they are related.

Contextual says “put this ad for a car on this page because the page mentions cars”. That is an $8 billion annual business for AdSense (gross simplification alert).

Behavioral says “put this ad for a car on this page even thought it is about sports because a few clicks ago this user was looking at a site with car reviews”. There are many successful ventures doing this.

Semantic technology can improve the relevance of both but is different. Here is how Wikipedia defines “semantic advertising”:

“Any technology based upon semantics, should be capable of identifying the meaning and context of the words on the page, in order to determine the appropriate advertising. This is an advancement to the current contextual model where the simple identification of keywords is seen as sufficient to represent the context of the entire page.

Semantic advertising is based upon two genres, the pure semantic solutions using ontologies and taxonomies, and those using technologies such as natural language processing and machine learning.”

Where Have All The Semantic Advertising Startups Gone To?

Wikipedia has an entry on semantic advertising. A week or so ago this listed about 10 startups, but some had gone out of business and many had changed focus to something totally different (or were never really semantic).

Wikipedia changes fast. As of today (May 6th), it only lists 1 semantic advertising venture which is Site Screen.

Somebody needs to edit that again as that is too restrictive. We certainly see Peer39 in that category and have covered them on this blog in the past.

The Peer39 Wikipedia entry does have a useful link to competitors:

* vexigo: uses NLP, does not look very active.
* iSense: aka SiteScreen, an active venture with a differentiated position.
* lucidmedia: an active contextual ad network, but seems a stretch to call this semantic.
* Hapax: site not found, presumably out of business
* 4adnetworks: looks out of business, domain parked
* Leiki: seems active, but contextual and behavioral not semantic as far as we can see.
* ad pepper media: the company behind iSense and SiteScreen (so not really another service)
* Proximic: appears to be contextual technology not semantic.
* Collective Media/Personifi: seems to be focused on audience analytics, not semantic
* Maxifier: focused on real time ad inventory optimization, not semantic.

So, it does look like only two ventures – Peer39 and iSense – that can be defined as semantic advertising on any reasonably tight definition.

Brand Safety For Exchanges & Marketplaces

Both Peer39 and iSense focus on the brand safety issue, even though their technology is different.

Sentiment analysis has been added to the old black-list approach. We obviously don’t want our steak knives associated with news about a brutal stabbing. But what about an ad that is associated with something that was very positive until all of a sudden it was not positive – such as Tiger Woods?

Sentiment analysis, which we cover as a separate topic, can clearly help with that issue.

Brand safety is a real issue for advertisers, a good problem to solve. But this is not a game-changing proposition.

What About The Other 5 Challenges?

In Part 1, we defined 6 challenges:

1. Better relevance to enable better matching. Semantic tech startups can demonstrate better relevance, but have not shown how to convince publishers, searchers or advertisers on any scale. Relevance alone is not enough to crack the market. Google is gradually improving relevance anyway.

2. Finding the user just in time to catch their intention. This is where Facebook may have a game-changer with their distributed Like button. That is an almost friction-free way to generate recommendations and they have enough user attention to use those recommendations to influence buying behavior. Who else can do this?

3. Brand safety. A valuable service to advertisers but not a game-changer for publishers or users.

4. Ad blindness. The only cure will be a huge increase in relevance. But this has to be done within the context of privacy (not “spooking the user” through really powerful behavioral tracking).

5. Getting recommendations without hitting the privacy wall. Facebook may be able to do this. Google, Microsoft and any other big web player would love to be able to do this. Aggregating recommendations is one thing, but doing it in a way that is a win for users, publishers and advertisers is still far from proven. The idea of a big data center controlled by a profit-hungry corporation knowing everything about individuals upsets enough people that regulators are jumping in even before market forces come up with a better solution.

6. Share of the cake. This can be bigger payout to publishers, lower prices for advertiser or an incentive to searchers/buyers or some combination of those. This can only come from a massive decrease in searching costs to enable more competition (in the form of lower prices to advertisers and/or bigger payout to publishers).

Potentially Disruptive Technologies

This is early, “bleeding edge” stuff. These are “straws in the wind” that indicate that change may be coming but that may not happen or not happen in any reasonable time frame. But there is new momentum on all these fronts:

1. A new “search stack” that will enable thousands of niche search services to compete with Google.

2. Bottom up tagging motivated by SEO.

3. User control over online identity.

A new “search stack” that will enable thousands of niche search services to compete with Google

Microsoft/Bing may chip away at the edges of Google’s market share (and that will be worth a lot of money) but it is unlikely that we will see a head-on challenger. The primary reason is that Google has approximately 1 million servers costing $ billions. Nobody will invest that kind of money to mount a challenge when all challengers to date have failed.

It is much more likely that the Google Leviathan will get tied down by thousands of Lilliputians that leverage a new “search stack”.

There are two components of this stack emerging:

a) distributed crawling through services such as 80 Legs or P2P search engines such as Wowd and Faroo. This is like Skype going up against Telcos with their huge investments in switches.

b) an open source software stack, particularly Lucene and Solr on Apache. The large amount of semantic web R&D going in universities will add to this stack.

The thousands of niche services will leverage this stack plus their domain expertise which maybe codified in the form of ontologies or domain-specific user experience design.

Bottom up tagging motivated by SEO

If the momentum to tag using RDF continues, it will be much easier for niche search services to do meaningful semantic search. That will be a game-changer. But it is unclear what is the critical mass that will create that tipping point and also when we will reach that critical mass.

But the momentum around tagging in RDF is strong and that will make it easier for Google and Bing. But more importantly it will make it easier for thousands of niche/domain-specific search services.

User control over online identity

If the movement to personally controlled online identities gathers momentum, the end user will “pull” vendors as privately as they want. This will be the most disruptive to current services. But it is also the most speculative and early stage currently.

SemTechBiz is Less Than 2 Weeks Away

The Semantic Tech & Business Conference (SemTechBiz) is coming to San Francisco on June 3-7! Join us for case studies, innovative panels, tutorials, and keynotes that will provide you with practical advice, hands-on guidance, and breakthrough approaches to solving business problems with semantic technology. Passes go up $200 at the door. Sign up now and save !