Friday, June 20, 2008

3 Ways to Scale an Advertising Business

I thought I’d try to tackle the “how to obtain scale in advertising” from more of a product development / product strategy standpoint. This is pretty basic though, so probably more geared for folks that haven’t spent a whole lot of time in advertising.

How to Get Scale

This is (perhaps) obvious, but I think there are three basic ways to get scale:
1. Build it
2. Buy it
3. Borrow it
Although not mutually exclusive, companies need to choose one of the three to focus on in terms of competitive differentiation. Each direction requires a slightly different mix of products and has a slightly different customer base.

Build – Focus on attracting your own audience. (e.g. build a media business like CNN, WSJ)
Buy – Aggregate or acquire audience by buying on the market either directly or through rev share (e.g. Ad Networks)
Borrow – Aggregate audience by “borrowing” or trading for traffic with sites that already have audience through building features or providing technology. (e.g. Build Facebook Apps, or provide search technology to Yahoo and AOL a la Google and Overture)

The "correct" answer on how you should attack the scale problem will depend on a host of factors including the evolution of the advertising ecosystem in your industry (i.e. dynamics are different if you are in mobile vs. internet vs. tv), your competitive advantage, and the relative strength of the other players in the space (in typical 5 forces fashion). Your decision should impact how your product development team prioritizes its product queue.

Building your own audience is probably the riskiest route but potentially gives you the most option value. Owning an audience lets you expand into “Buy” or “Borrow”, but the reverse is not as easy. If this is your focus, you should be heavily weighted towards consumer-centric features and products.

Buying audience requires a large existing market and a pre-existing network of buyers and sellers (doesn’t quite exist on mobile yet for example). It also requires a larger budget, but if your analytics are good enough the risk isn’t actually that bad (there are a surprising number of ways to arbitrage ads). If this is your focus you should be heavily weighted towards metrics, analytics, and integration with multiple parties in order to find and exploit market inefficiencies quickly.

Borrowing audience requires some technology advantage or asset on your part to exchange with the large players that have audience (like the carriers). Working with large players also requires longer sales cycles so you may be looking at a deeper and longer “J-curve” than for other routes. If this is your focus, you should be heavily weighted towards building technologies that integrate easily with the top players.

Again, these three paths aren’t mutually exclusive, but if you look at your development resources and they are split evenly between all three, you are most likely unfocused and spread too thin. In addition, the investment criteria and early success metrics tend to be different depending on which way you attack the scale problem. My preference is to choose one direction and try to knock it out of the park, although its certainly possible to do two (like Google). In those cases, though, their “Borrowing” technology is usually an easy offshoot of their consumer “Build” site.

A big part of your decision will be around the type of audience you want to try and aggregate. As Andrew Chen has noted, CPM values will vary a lot based on who is using your site, where they are located, and whether or not it’s a niche focus. The minimum scale you need in terms of pageviews is inversely-related to your CPM, so for modeling purposes spend the time to get a realistic CPM value for your audience. Once you have a realistic CPM, you can figure out the minimum scale you need with some simple math:

Just take your monthly target burn rate (let's say $1M) divide it by your expected CPM and multiply by 1,000

So let's assume your burn is $1M
Let's assume you figure out your CPM is going to be around $5

$1,000,000 / $5 CPM * 1,000 = 200 M Impressions / Month

Then your minimum scale for becoming an "ad-supported" business is around 200M impressions per month.

There are a couple of things to think about when you calculate your minimum scale and start thinking about your likely advertisers:
  1. The more unstable your source of impressions the more unlikely you are going to get the big budgets. Advertisers (whether they are direct response or not) still need to plan their spend, which means if they allocate $30k to you they want to make sure you use it at the ROI they expect.
  2. The smaller the average budget the more automation you will need (and your development costs might go up increasing your burn, and consequently raising your minimum scale number)
  3. There is generally such a thing as a minimum budget for advertisers. For most companies spending less than $5k / month on your site may not be worth their time - unless you happen to be aggregating a very unique audience.
I'm skipping over a few things, but at the very least at the end of this exercise you have a better idea of what "scale" actually means relative to your idea, your company, and your industry. [As a total aside, I believe that most folks understand that advertising needs scale, but most don't really grasp the magnitude. It’s sort of like the fact that there are more stars in the universe than individual grains of sand on the earth (some math on this here) or the fact that life on earth has been evolving for potentially 3-4 billion years. One can hear the number, but the entire concept of a “billion” anything is a hard one to get one’s head around. It’s a similar thing when someone says, “Google served 5.9 billion queries in February”. The scale is so massive its almost incomprehensible and meaningless - especially if you are just a small startup trying to make $1, let alone $1,000,000,000. Jeremy Liew covers some of the math here and Mike Speiser correctly points out "The legions of entrepreneurs out there should appreciate just how MASSIVE you need to come in order to build a big advertising-based business.”]

One thing to watch out for: when you are doing your modeling don’t make the mistake of thinking that your ad technology can change the inherent value of the audience. I have yet to see any ad technology alchemy that can make online lead (broad social networking) into online gold (search) (although there is technology that lets you find the gold among the lead). So, if you have assumptions in your model that say "the social networking avg. CPM is $0.30, but my average will be $0.60 because of technology" - well, I would be wary of that. (More to come on this topic later)

A lot of how you decide to split your resources depends upon where you think your competitive advantage will lay. For example, if your company DNA is that of math PhDs then I would seriously think about focusing more on building fancy ad targeting and analytics and taking a "Buy" strategy. However, if your company DNA is of folks from consumer products then it would probably be a mistake to invest a lot of resources in building your own ad serving / ad targeting technology and better for you just to double down on building audience. Again, this seems obvious, but I think really smart people sometimes have a tendency to make simple things more complex than they really are. It's sometimes surprisingly hard to keep things simple.

Once you have picked the audience you are trying to aggregate and have an idea of the advertiser base that would buy that audience a lot of the tactical questions start becoming more obvious. Your target audience becomes more clear, your product prioritization becomes more clear, and most importantly your value proposition becomes more clear (both to your audience and to your advertisers). You know exactly who you have to reach (either customers, publishers, or partners) and therefore you can figure out what they need. As you develop your product to meet those needs, you’ll be well on your way to achieving scale.