Your browser is not supported. Please update it.

1 March 2006

Jostling amidst the great IPTV advertising rethink

We go on rather a lot about advertising in Faultline, mostly pointing to the inevitable harm to traditional TV advertising models as the video entertainment market fragments into multiple layers, many of which require the funding of advertising.

This spreads a slightly increasing amount of advertising spend over a massively increasing number of options. The result is many more TV services, each with increasingly smaller slices of a pretty much the same sized cake.

Global advertising spend rises each year by roughly 5% to 6% and according to Zenith Optimedia reached $403 billion during 2005, with some 37% of this spent on TV advertising. The US accounts for roughly 42% of the total ad spend, and by implication the US TV ad spend is roughly $60 or $62 billion.

For content companies and channel owners it doesn’t matter, to a large extent, whether this advertising is delivered by broadcast, cable, satellite or IPTV, given that a TV channel owner will have deals with each of these routes to market, and as they wax and wane, a TV channel’s ad appeal varies less with the method of delivery and more with its underlying popularity.

But once you add in mobile TV, DVD viewing, and more time spent playing games and browsing on the internet, then collectively ALL of these types of major TV platforms are delivering collectively less viewers, and in the end this must tell on the advertising appeal of each channel.

For the underlying networks it will show in the carriage charges they pay, which are privately negotiated based on how many viewers the operator can deliver to the TV channel.

But broadly if satellite continues to make up an increasing number of US viewers, then its buying power goes up. Its cost per viewer goes down. This will mean that any TV operator that is delivering less viewers will end up paying more per viewer. At the moment we are seeing this in a gradual shift from prosperity in broadcast networks to cable networks. But soon this shift will be evident from these into other types of TV delivery platforms.

So it was nice to have the opportunity this week to talk over this scenario, with a company that lives and breathes advertising on IP systems and talk about the types of weapons that might come to bear to make advertising in some of these markets, especially on IPTV systems, more effective.

Patrick Christian is the founder of 18 month old start up Packet Vision, which last week raised its second round of $1.8 million of Angel funding.

Back at the IBC shown in Amsterdam in September, Packet Vision was one of the few companies offering anything that addressed the advertising richness promised in IPTV networks.

It offers a hardware hub that sits inside the IP network, typically it might be installed close to the encryption server at the head end, which is a multi-processor architecture, based around multiple MIPS chips. It acts as an advertising server, splicer, router and management system for delivering separately targeted advertisements to demographically different audiences within IPTV networks.

‘Our approach is to have all of these functions in one slimline box with gigabit Ethernet the only way in and out, more like an internet appliance. Networks can begin with regional delivery and then gradually distinguish by age groups and get more and more accurate demographic groupings, installing appliances as they go.

‘The existing cable network advertising insertion companies may offer a lot more services such as VoD, but they require huge 7 foot racks of equipment and to do this they would need huge mainframe like installations.’

The basis of Packet Vision’s claim to fame is that IPTV operators can now personalize TV advertising. ‘When you watch the TV, during a 3 minute ad break there is probably only one of the ads that is showing anything that you are interested in. Personalized advertising is about trying to send more effective adverts to people based on demographic groupings,’ said Christian.

Packet Vision has built it to work with standard IGMP multicasts or with unicasts, and it runs under the control of a Linux server with other servers acting as a network of video switching engines. Third party systems holding demographic data can be integrated from middleware held data or from external packages.

At the moment Christian is trying to retrofit the system onto existing IPTV environments and says that although he’s had conversations with some of the tier 1 telcos on each side of the Atlantic, the next stage for his company is to actually install the completed architecture into a network and find out what the real world gremlins are. He believes that this kind of real world experience is what will keep Packet Vision ahead of all the IPTV advertising companies that are bound to enter this space once it really takes off in anger, and he has two installations going in, one each side of the Atlantic, that will provide this. It will also provide future investors with the kind of confidence to invest in subsequent venture capital rounds, by establishing a higher valuation for the company.

‘Advertising is not yet ready, not even IPTV ads, to service a different advert to each consumer. It is more likely that TV advertising can take more localized advertising and offer more localized services, so that geographically close neighbourhoods can be offered local ads.

But the technology has to work quite hard to do even that. There are three or four server companies that offer ad insertion in the US to the cable community, primarily Seachange International, C-Cor and Concurrent, usually incorporating separate equipment for automatically splicing video together, such as from Terayon or Big Band networks.

Neither cable nor IPTV networks will allow content that has just been re-compressed, and in IPTV’s case IP packetized by the head end, to go anywhere until it has been encrypted. So ad insertion engines all have the problem of dealing with encrypted content.

‘We do our insertion at the MPEG transport layer, which makes it quite difficult if the entire stream has been encrypted. But these days at least the MPEG headers are in the clear, which means that we can insert adverts without decrypting and then re-encrypting the video stream,’ said Christian.

This would make a bit of a mockery of going to all that trouble to encrypt content in the first place if you were going to have to have the horsepower in the next box, the ad insertion box, to do the job all over again. With modern encryption which operates around precise crypto-periods of usually 10 seconds, the entire stream would have to be re-encrypted, because the advertising parts would shift the timing of the stream so that the wrong key would be applied to it otherwise at the set top end. Or at least some kind of timing interval introduced to inform the set top.

In less rigorous encryption environments such as cable and satellite, these use older and less advanced encryptions systems, although they do offer secure hardware keys in smart cards and tamper proof chips. But they also have developed habits of changing the encryption keys less often, like every program, rather than every ten seconds or so. This means that it is probably an easier trick to build a digital ad insertion technology for these, than for IPTV.

Currently advertising is always valued at a particular price for each 1,000 impressions that are delivered. This is more expensive if it is perceived that there is more chance that a consumer will, or can, act on it, so naturally just introducing local advertising makes this rate go up.

And IPTV ad insertion systems are all about countering the arguments that we raised at the beginning of this piece, about advertising reaching less eyeballs, being spread over a greater number of more fragmented services.

Today the actual numbers of viewers of a particular advert are calculated, in broadcast networks, based on interviews with viewers and statistics compiled on things like people meters (devices that a representative sample are supposed to click each time they change channel). These are compiled by such esteemed research organizations as Nielsen Media, which has a virtual monopoly for this in the US.

In networks that have a return path, all of this guesswork goes out of the equation and each time an advert is viewed, or a channel is changed, especially when a channel is changed when an advert is in progress, then that data is carried on the IGMP stream back to the advertising management system and to any other piece of middleware that has a right to view it on the system.

‘There is a lot of research,’ says Christian, ‘that says that it takes only 4 viewings of an advert to get the message across to each viewer, but that after 30 viewings, the viewer is now irritated and beginning to feel bad about the product .’

‘IPTV promises one day to intelligently ‘frequency cap’ ads so that no home can be subjected to them more than 30 times. This of course raises a picture of some homes in a neighbourhood having seen their 30 iterations, and which now need a new ad to be served in its place. While other homes, either because they were watching other channels or because they watch less TV, or were in the wrong demographic, are still good to see the original advert.’

This will mean increasing the number of separate multicast streams that exist on the IPTV network, one with one ad, and another with alternatives. In order to do this, it would mean installing more of the Packet Vision Video Switching Engines, but the result would mean that an advertising campaign can be successfully delivered in less ‘impressions,’ to borrow language from the internet advertising world.

We get the image of a campaign of progressive ads which engage the consumer in a story. When they are shown on existing broadcast networks, the first one is shown until everyone is sure to have seen it, and then the entire broadcast group moves on to the next one. In the UK there as a famous story of a couple courting and teasing one another through their sharing of a famous coffee brand, which took years to unravel.

Eventually in IPTV networks each advert can be shown 2 or 3 times per household and then the next in the sequence of adverts could be shown, until the entire message had been delivered to the TV household. Other households might operate at a different rate. As Christian says, IPTV is not on the verge of making that a reality, a separate ad per household, not quite yet, but it is these kinds of models that will help TV networks retain high ad rates and therefore high stock valuations, at a time when the TV market is further fragmented by video downloads, podcasts, web TV and mobile TV services.

‘Using this approach of only sending adverts to demographic groups that are relevant to the ad, operators can sell the same time slot 6 times over. It is more targeted, so there is less wasted bandwidth sending adverts to the wrong age group or the wrong sex or economic grouping. Sure it is selling the same ad six times to one sixth of the audience, but each time it is worth more because the advert is more likely to work.’

This will eventually turn TV advertising into something more like internet advertising with an advertiser able to pre-book a set number of ad impressions to qualified demographics only.

So the winner in the technology stakes to lead that revolution, will in turn retain their own high valuation, and there is a clear and unambiguous way that this can be achieved.

IPTV operators will not want to spend large quantities of cash on network resident devices that simply allow them to serve different groups different ads, unless it increases their revenues. It makes sense to ‘vendor finance’ these devices (have a finance company pay for them) and for companies like Packet Vision to take a percentage of the increase in advertising revenue.

If an advertising innovator generates lots of revenue improvements for a number of IPTV operators, then this will result in extremely high revenues for the system vendor and the potential for a huge exit route.

At present, looking at the US market, SeaChange International and C-Cor are public with respective valuations of $270 million, and $345 million on revenues of $122 million and $246 million. Of course not all of this is advertising insertion revenue, and it includes VoD services, but then again the same applies to inserting advertising into pre-encrypted VoD streams.

Another issue that Faultline sees as becoming very big is retrofitting the same technology to cable operators. ‘We are sticking entirely to IPTV, but we acknowledge that cable has many of the same elements and we would consider licensing our technology to another company that wanted to attack the existing cable market with serving personalized advertising to cable operator customers,’ confirmed Christian.

What about the ability to experiment with new types of advertising such as viewing advertising voluntarily on a DVR. This is something that TiVo has pioneered very successfully, with 95% of the ads that it downloads to its DVRs, being viewed.

‘We could accommodate that type of experimentation. We want to keep our device in the network so that we can deliver adverts anywhere they are needed. If you want to download them to key DVRs, that’s fine. If you want to offer links to the adverts addresses on an EPG, and view them straight from our server, that would also be fine.

‘But the advantage of holding them in the network is that if you suddenly want to launch something like a mobile TV service, or partner with one, and run the ads into the Mobile TV service, then this is a relatively trivial change.’

The rapid introduction, this year and next, of IP based mobile TV services, of course makes any business innovating in IP delivered advertising begin to look even more attractive. Datacast services such as DVB-H and T-DMB don’t really gain anything from a system like this because they’re in essence broadcast, but multicasting cellular delivery, offered by companies like IP Wireless in its recently launched TDtv system, could deliver one set of ads to one audience and different ads to another. The only disadvantage is that each new advert feed needs a separate multicast channel and in mobile right now there isn’t too much bandwidth available for that. Of course more to the point each base station might deliver a different set of local ads ‘ that wouldn’t take up any more bandwidth, at least not in the air interface.

While we are excited by the prospect of this type of function unravelling over the next five years or so, it will be a rocky ride for companies like Packet Vision, as all the established cable players try to make a play for IPTV advertising systems, and it may also be a natural place for middleware operations to play.

But if IPTV networks don’t put services like this in place early on, they will risk offering too small a footprint, and one too similar to cable, and will suffer from the comparison.