A VC - Latest Comments in Scale Economics

Re: Scale Economics

fredwilson — Wed, 28 Jan 2009 19:14:21 -0000

It would be interesting to hear more about what is easy to scale and what is
not

Re: Scale Economics

Michael Stephanblome — Wed, 28 Jan 2009 15:27:57 -0000

Thanks Fred. I like your blog - lots of interesting observations. I started an online display exchange about two years ago in Europe and had to learn the hard way in which areas you can bring scale and efficiency through technology to advertisers and publishers and in which area the current intermediaries are so powerful that one startup alone cannot overcome them just now. So I am still hopeful that more transparency will eventually lead to more efficient pricing and hence value creation but you have to take it one step at the time.

Re: Scale Economics

fredwilson — Fri, 23 Jan 2009 20:59:18 -0000

Great comment

Sounds like you know a lot about this sector

Re: Scale Economics

Michael Stephanblome — Fri, 23 Jan 2009 19:02:41 -0000

I think that the economic crisis could be a catalyst for a surge in online performance display advertising. The lack of online brand advertising budgets will force online publishers to dedicate much more thought on how to effectively monetize their remnant inventories. They might quickly realize that the lack of transparency from blind ad networks is detrimental to their eCPM and that they need to look at ad exchanges and understand how optimization works.

On the ad buyer side many media agencies are realizing that they need to become smarter by using technology (which is seldom in their current DNA). They are used to outsource their display performance budgets to blind ad networks and that has made them vulnerable in terms of margin and USP. Clients are scrutinizing every penny and are asking what their agency is doing about it. That's why we are seeing some new types of media exchange agencies emerging who are trying to build audience platforms who will enable them to find relevant user cookies for their customers on the exchange in order to reduce wastage and increase ROI.

I think 2009 is too early. The so-called ad exchanges are still in their infancy and online display advertising itself still has many hurdles to overcome. Automation is happening slowly - self serve is only working for very low-value inventories and advertisers. But it will come - the whole industry has to become more efficient and technologies for targeting, optimization and efficient creative production are slowly coming together.

Re: Scale Economics

Jonathan Wegener — Fri, 23 Jan 2009 01:43:08 -0000

Yes rtoennis! Couldn't agree more. Glad someone finally said what I was thinking.

CPM/display is shouting -- it's pushing a message onto unsuspecting victims who aren't open to hearing that message.
Search advertising works because the ad is displayed at the moment where the visitor is SEARCHING for something specific. It's the perfect place to display an ad and the user is generally receptive to it: thus the order of magnitude higher click through rate on a SERP page vs a generic web page.

Re: Scale Economics

fredwilson — Thu, 15 Jan 2009 08:22:44 -0000

Yes, search did spoil marketers

But therein lies an opportunity to build tools that deliver the analytics
and targeting that can show them how to use display effectively

Re: Scale Economics

Gui Ambros — Wed, 14 Jan 2009 22:47:14 -0000

good post.

Have also a quick interesting story to share:

a few years ago I conducted a study between an automotive client and one large portal/ISP in which we were investing a couple of million dollars per year. The study consisted in matching the ISP's customer database with the company's database of new vehicles sold in the past 6 months. We then analyzed the number of times each user saw a banner, and how many times it clicked.

As you may have expected, the result was surprising for us and for the client. We proved there was a STRONG (and statistically significant) correlation between # times a banner was shown to a specific user and the % of users that bought a vehicle. *Regardless* of click.

(of course users that clicked had a higher conversion rate, but this was expected)

Having said that, the problem most companies - and vendors - are simply unable to track and measure impact cross-channel. And even looking just at online purchases, advertisers still struggle to identify impact of display impressions over conversion rate on search, or the spillover on the brand. If you add call center, retail stores, etc to the mix the confusion just increases. Agencies and Clients that use DynamicLogic studies are even more scarce.

You know what's funny? Search spoiled marketers. We don't use the same scrutiny to measure print advertising, radio or broadcast, but online brings this stigma that "everything is trackable", so in the meantime let's bang display on the head...

Re: Scale Economics

fredwilson — Tue, 13 Jan 2009 18:43:18 -0000

Keep working on it, we need display to work better for advertisers with so
much of it hitting the market this year

Re: Scale Economics

davidhelmreich — Tue, 13 Jan 2009 15:55:44 -0000

Fred and Brian -

All good points, and I think many of the individuals on here would be ecstatic (and simultaneously mortified) if display can ultimately outperform search. It would dramatically change the lead generation industry and finally threaten the large search giants.

My experience with the networks is similar to Brian's in that they are technology companies first, second, third, fourth, and maybe marketers and media optimizers somewhere much further down the the list. This has created a unique opportunity for companies like MediaMath, Invite Media, mediageeks, and others, than are fighting to show the value of publisher's audiences, and make it easy to agencies and networks to find individuals across mutliple sites, rather than buying on a mediakit basis.

Today, it still seems as though the art and audience segmentation is largely done offline by the agencies and advertisers, and then the target audience is sought through a mangled insertion order and translation process that essentially flattens the ideal lift curve down to a straight line. There isn't an easy way for an agency or advertiser to easily find their target audience online, period. By pushing this square peg through a round hole, the networks are forced to "interpret" the desires of the agency, and in more cases than not, just do what they want, and report something after the 30M impressions are fulfilled.

Behavioral targeting, although it may employ the greatest technology and clickstream translation layer, resulting in some behavioral segment, isn't understood by media buyers. We sit on the sidelines and remark how neat it is, but the sales teams and advertisers can't find a relevant audience that way.

Depressed CPM's is only excaberate the problem as more media will be purchased and more RON impressions will be used to backfill large orders.

In full disclosure, I have been working for over a year to combat this problem and provide a solution to power ad networks to drive informed ad serving and reach the true audience requested, but time and time again, networks want to tout their own capabilities and fall back on science and optimization, which in many cases, in neither science, nor optimization.

"The Ad Server just figures it all out in real-time" won't work in 2009 and 2010.

-Dave

Re: Scale Economics

vijaycs42 — Mon, 12 Jan 2009 18:59:54 -0000

I agree that display cpms will drop, and had done a blog post in August on this subject: http://clickr.typepad.com/v...

I also agree that some forms of targeting will become more commodity. But more than that I feel display ad targeting is highly overrated. Most display advertisers have a goal that's closer to brand advertising than to getting immediate conversion, and for that goal they would not need more than some basic forms of targeting (e.g., high level subject area, location etc.). Beyond this basic level, you could look at any metric to figure out that more advanced targeting doesn't help in either click-throughs, conversions, or awareness. If advertisers have more specific goals in mind, they'd be better served by search advertising.

Re: Scale Economics

fredwilson — Mon, 12 Jan 2009 16:51:35 -0000

Great post. Using keywords to target display "intelligently" is a great development.

I'll try to see your panel. Scott is indeed venerable.

Re: Scale Economics

jonathanmendez — Mon, 12 Jan 2009 14:55:44 -0000

The pre-click and post click technology and strategies of search are certainly moving into display thanks to APIs and semantic tech. Performance is moving along with it. "The time is right for a palace revolution."

You might be interested in a post I did two months ago on this http://bit.ly/OWYx. Also in March I will be moderating and speaking on the first conference discussion on this sea change (joined by the venerable Mr. Rafer) at Search Engine Strategies, New York.

Re: Scale Economics

fredwilson — Mon, 12 Jan 2009 11:42:26 -0000

Here's a post I did on some comscore research that does look at the longer term affects of display

http://www.avc.com/a_vc/200...

Re: Scale Economics

Richard — Mon, 12 Jan 2009 11:27:16 -0000

We at Mediashakers.com are seeing more and more advertisers as well as smallmedium sized new advertising that are coming to take advantage of the mass of inventory available at great rates and ona wide range of site. We are also seeing many more publishers expressing their willingness to allow us to manage their entire invetory.

Re: Scale Economics

BmoreWire — Mon, 12 Jan 2009 11:12:05 -0000

Well, I think display is at a completely different stage in the purchase consideration funnel and the problem is that people are considering them equal. To say that display ads in social networks don't work, is not necessarily correct. As you drive down Interstate 95 and speed by outdoor ads for BMW and then you purchase a BMW, just because you don't use the 800 number on the outdoor ad for that dealership, doesn't necessarily mean that after years of exposure, BMW's display advertising had no affect on your purchase. A big problem is that people evaluate online display in the short term. I would love to see research or a study of a major online display spender's long term effect from heavy banner advertising.

Re: Scale Economics

fredwilson — Mon, 12 Jan 2009 08:17:10 -0000

Thanks for stopping by and weighing in scott. You are way closer to this
than me. We¹ll see but I wouldn¹t bet against the exchanges.

Re: Scale Economics

Gui Ambros — Sun, 11 Jan 2009 23:52:27 -0000

Good comments, Fred and Brian. Though, I believe we're far from having Display surpassing Search.

Costs are going down because inventory is going to the sky. Portals and publishers are adding social content (e.g. Yahoo, MSN, Terra, etc), which is good to increase traffic. But the traditional advertising model for display - interruption marketing - don't work for social content at all. Users don't like to be interrupted when they're poking people, sending vampire bites, reading emails or watching videos of a dog peeing on your pants. It's simply human nature. As a consequence CTR have been going down, and will continue so in 2009.

Smart marketers are using rich-media and good creative to compensate, but as Brian mentioned, this is art, not science. It's a trial-and-error process, and there's simply no way to be always right. Sometimes you get it right and have great CTR results, sometimes not. Adaptative Ad Serving techniques (Dart, Atlas, etc) help a lot to optimize testing procedures and thus increase CTR, but at a high cost (specially with pennies CPM and you having to serve billions of banners for a single campaign).

My prediction: things will continue flat display & search in 2009. Display Media CPM *and* CTR will continue going down, thus keeping CPC flat. Search CPC will increase a bit (we're seeing this already: bids are getting higher over time, with more competitors entering the game, specially outside US), but will still have a better ROAS than Display.

BUT... things is not all rosy to google. Marketers are realizing Google is like the salesman that see the customer going to the checkout and stamps his code, so he can get credit for the commission. Of course Search results are always great: the hard work of getting the customer to buy was done by the TV, Print, Display, Social Media, bloggers, twitter, retail store, etc. Lucky GOOG that most companies don't have the tools or desire to measure properly this spillover and calculate the REAL customer acquisition cost... Ignorance is a gift to some :o)

(disclaimer: I work on a marketing agency, so demand generation campaigns are part of may day-to-day)

Re: Scale Economics

rafer — Sun, 11 Jan 2009 22:54:50 -0000

Sorry, Fred. Have to completely dissent on this one.
1. Buyer aggregation is going to kill publisher aggregation. http://www.marketingvox.com...
2. The best new source of display targeting is search KW retargeting, which performs at ~20% of search (which is better than contextual).
3. Publishers without focused or easily segmentable audiences are generally, and maybe permanently, screwed.
4. Exchanges won't survive the recession in their current form. The buyers are in a position to insist on flat-fee technology service suppliers and there's enough mediocre inventory to erase any scarcity of demand as #1 on this list becomes increasingly automated.

Re: Scale Economics

fredwilson — Sun, 11 Jan 2009 20:12:20 -0000

Not likely

Re: Scale Economics

BmoreWire — Sun, 11 Jan 2009 16:34:40 -0000

You're probably giving my opinion too much credit but thank you, and I will continue to offer commentary.

Re: Scale Economics

fredwilson — Sun, 11 Jan 2009 09:10:07 -0000

Brian interesting that you¹ve been at the center of both posts in the past
couple days. You clearly are working at the epicenter of all of this stuff.
Thanks so much for all your comments. They are great.

Re: Scale Economics

fredwilson — Sun, 11 Jan 2009 09:09:10 -0000

Right, but it pays the bills for a lot of stuff we want and need

Re: Scale Economics

fredwilson — Sun, 11 Jan 2009 09:07:09 -0000

You said it better in many fewer words than I did Jordan

But that happens in the comments on a regular basis!

Re: Scale Economics

fredwilson — Sun, 11 Jan 2009 08:39:12 -0000

The vacant main floor of a store called ³apartment² in Berlin

The store is in the basement and for some reason they have this 3000sf main
floor retail space that is completely empty

You can see it on their web page http://www.apartmentberlin.de/

So I decided to turn it into a studio for 7 seconds with the help of Josh
(who makes his cameo at the end of Charlie¹s film)

Re: Scale Economics

Willan Johnson — Sat, 10 Jan 2009 22:59:21 -0000

Great comment rtoennis. Search advertising works because there is value for both the advertiser and the consumer. Hard to imagine display ads outperforming search without such value.

Fred, agree that CPMs will generally fall, and targeting will improve. But another challenge with your view is relating broad reach (i.e, millions of impressions) with penny CPMs with sophisticated targeting. If the targeting is so good, why are advertisers using exchanges to buy millions of impressions? (In other words why use a shotgun when a rifle / scope will get the job done more effectively).

My sense is that traditional display advertising will only become more effective through better targeting, stronger engagement with the consumer, and some form of direct benefit or feedback. Penny CPMs will be a part of the broader solution, but not the core.