Joho the Blogonline ads Archives - Joho the Blog

January 12, 2010

[berkman] Fernando Bermejo on measuring advertising

Fernando Bermejo is giving a Berkman lunchtime talk titled “Mapping Online Advertising: From Anxiety to Method.” He says that people sometimes hear that he researches advertising and assume t)hat he works for advertisers and wants to improve advertising. In fact, his interests are scholarly. He’s going to talk about the dynamics and logic of online advertising. (The subtitle of his talk is a reference to Devereaux.)

NOTE: Live-blogging. Getting things wrong. Missing points. Omitting key information. Introducing artificial choppiness. Over-emphasizing small matters. Paraphrasing badly. Not running a spellpchecker. Mangling other people’s ideas and words. You are warned, people.

He begins with the John Wanamaker (1838-1922) quote: “Half of the money I spend on advertising is wastsed. The trougle is, I don’t know which half.”: It’s the most repeated sentence in advertising, although FB notes that it might be apocryphal. It manifests anxiety. In 1923, people were urging advertising to become a science. From anxiety to method. And that repeats online where there’s anxiety about online ads failing, and a growing desire for a reliable method.

FB explains that anxiety is a free-floating fear with no particular trigger. In this case, anxiety arises from the growth of the gift economy (= peer production) and the pay model. The traditional ad model is advertisers <> media <> audiences, but that doesn’t capture all the complexity. We need to insert “products and services,” since the aim of the ad is to get people to purchase those products and services. Audiences may see the ad but not make the purchase. Hence anxiety.

The method, FB says, is: Collect info. Metrics. Targeting. Have the right context. Price. Funnel.

Collecting info: Online, every interaction can be recorded, but those are discrete events. Advertisers need to figure out ID or profile. Cookies are one approach. FB points out that most people don’t know that Flash sets cookies also. 54 of the top 100 sites use Flash cookies, but only 4 mention it in their privacy policy — Soltani et al, 2009. When you opt out of cookies on a set, your Flash cookies generally are preserved. Another study (Kryshnamurthy & Willis 2009): “Our results show increasing aggregation of user-related data by a steadily decrteasing number of entities.”

Metrics: It takes an idea from the press (page) and broadcast (time) and calls it a “visit.” This seems to FB not to be very apt since you’re not on a site continuously necessarily. People also measure “engagement,” but that’s very hard to measure. He shows a complex mathematical function that seems somewhat silly.

Targeting: It has gone from audiences to highly granular “targets.” Advertisers have used demographics for a long time, as well as looking at the content consumed. But ther’s also content produced (e.g., search keywords) and behavior.

Having the right context: Previously, an advertiser was confident the context of the ad wouldn’t be weird because the TV station of magazine was professional, etc. . Now, online advertisers produce content specifically for advertising (e.g., Demand Media), and want to control user-generated content.

Pricing: It used to be CPM (cost per million). Online, CPM is 38%, but performance-based pricing is 58%, including cost per click or cost per lead/purchase. Obviously, the content producers don’t like this because you can produce audiences to sell, and then it’s up to the advertiser to do an appealing ad.

Funneling: You can show people your ads, but you can’t make them take action. There are ways to funnel people toward a purchase. Branding is important because the worst result of an ad is that it stimulates a purchase but of a competitor’s product. There’s also tracking and re-targeting: If a visitor sees your ad but goes to another site, they’re shown an ad reminding them to buy. Another movement: Funneling used to be just about impossible: After showing you the ad, the advertiser had no idea what you went and did. Now customers can be put into more controllable environments, especially the mobile Net. (Google and Apple have both bost companies in the “mobile intelligence” area, FB says.)

The consequences of these are diffuse. Collecting info means that companies want to collect more and more info. Metrics: Advertisers are uncomfortable with that which cannot be precisely measured, so they prefer search engines to social media. Generating the context means that the advertisers prefers certain environments and shapes context. PRice: Adverteiser tries to reduce risk by asking others to assume it. Funneling: Adverister tires to be more intensive and extended. [Missed one. Sorry.]

Broader consequences: Privacy issues makes advertisers play a cat and mouse game. Search engine advertising disintermediates the content producers, and to generate behavioral info the content producers need to provide their info to cross-company entities. [I missed his point about structure.]

The online content layer can only be explained through the logic of gift, of pay, and of advertising. We need to learn more about advertising

Q: [doc] Is advertising a bubble? Are we too dependent on Google which is 100% dependent on advertising?
A: So long as we think advertising has an effect, it has consequences.

Q: [carolina] Are we back to anxiety?
A: Yes. After all the method, we’re back to anxiety. There’s always a source of anxiety because the final goal is perfect control, and that’s not possible. [Note: FB is not advocating perfect control!]

Q: There are networks of ads, with many hops between the site and the advertiser. That makes tracking even harder. Why do we demand less anxiety online? Why not just put your message out there?
A: Because there’s the possibility of having more accountability. The audience measurement industry is somewhere between making it up and measuring it exactly. We have a need to believe there’s something tangible and real there.

Q: The old way of measuring success was just to put your ad on TV and see if sales increase. Does that still work?
A: There was never proof that advertising works. It may shift people from one brand to another but not increase consumption. The online evidence isn’t very clear or reliable.

Q: E.g., affiliate marketing offloads the risk onto the affiliates. Not everyone who embraces risk is anxious.

Q: Ads are not under the announced privacy policy of the site they’re on.

Q: [me] Couldn’t we say that the new models are the end of anxiety about advertising? You know what you’re paying for (pay per click, etc.) and there’s a market mechanism (auctions) that leads toward semi-rational pricing.
A: You still have to worry that people would have bought it anyway.

Q: [salil] If there’s a fixed budget for advertising, and if they optimize for, say, the number of clicks, then everything’s a science. But there are still two choices that are anxiety making: How much of the budget should be spent on ads? And how does that final metric translate into increase in revenue.
A: Yes. If people click on the ad mean that they’re more interested, does it mean you’ll get more purchases, etc.
Q: Long tail advertisers maybe should have less anxiety because they can more easily correlate ads and sales.
Q: [rob] And there are short term blips that can be tied to advertisements.
A: Yes. It can be very different for different types of products.

Q: [jason] Death of brands?
A: Context-dependent. Brands will continue to be important.
Q: [judith] Efficiency is great for consumers — we’ll just buy what we want to buy. But it seems like a huge part of the ad industry is aimed at creating desires. Is there a way of giving people a measure of how free the Web is not? How much are we paying for our content through unnecessary consumption?
A: These are core issues. Is advertising just about informing people or about stimulating desire? Those two things can co-exist.
Q: [judith] You can pay for content by getting people to buy stuff, but they won’t pay for the content they’re there to read.
A: I’ve tried to stay away from the culture impact of advertising in this talk. Perhaps another talk …

[Posted without re-reading.]