Wednesday, December 04, 2013

Blurred lines: really part 2 this time

Sorry for earlier posting difficulties. 

The Lessons of Nauru
Bob Garfield, Co-host of On the Media and MediaPost columnist

Nauru: was wealthy; phosphate resource from centuries of built up guano was valuable, but only took 50 years to destroy, and now poverty is endemic.

[came in on the middle] The advertorial: “Borrowed interest”: most advertisers have been willing to label, relying on format and proximity to real ads.  Never a really big revenue generator or ethical problem in the newspaper days.

That was then.  Atlantic/Scientology: the most incriminating example of media prostitution.  Actually less worrisome than far less lurid publisher-advertiser dalliances, because this was so over the top that it was instantly pounced on. The real danger is what’s undetected.  The Atlantic also ran an IBM ad “Why Social Media Matters for Your Business,” but consumer wouldn’t be likely to recognize it as an ad. Doesn’t the reader have the right to know whose interests are being served by the content? Media means in between; the whole point is to have a third party at arm’s length. If IBM is such an authoritative source, why the charade? Why not proudly slap their authoritative logo over everything? Well, then it would be an ad that no one would read; that’s the central truth that can’t be rationalized away. It’s all based on consumer confusion. IPG MediaLab/Sharethrough study found that study subjects were 25% more likely to look at a native ad than a banner; they’re proud of this!  Native ad isn’t merely a deception, it’s a conspiracy. Even worse: the fake link/clickbait headlines as if editorially generated.  Wrappers for get rich quick schemes. 

Maybe you think these are suckers, or that an entire industry shouldn’t be condemned for a few bad apples. But the stakes are higher.  Native ad from Forbes, one of the most aggressive but also straightforward—type mimics rest of publication, but brand voice logo is prominent and at least there’s an explanation if you care to click through.  “The New ‘4Ps’ of Social Business Marketing.” That content migrates.  Within 8 hours of hosting, it had also shown up on 162 sites, and as far as those were concerned the source was simply Forbes.  The internet doesn’t know you’re a dog, and it doesn’t know you’re an ad. And here we are talking about this as if it were some sort of a savior. 73% of online publishers association already accept native ads, and 17% more to come soon!  Even though as currently deployed it violates the most basic publishing ethics.

Why? Existential crisis is always a bull market for noble ends invoked to justify dubious means. Save journalism from destruction!  If the police are underfunded, we do not think that a good idea is to sell badges and uniforms to whoever wants to buy them, on the theory that it will produce revenue and some of the ‘cops’ might deter crime by virtue of hanging around.  Trust is not meant to be a barter item. If trappings of trust can be purchased, public is exposed to deception. WGN’s “High Dividend Stocks of 2013” aren’t really from WGN.  We aren’t being saved.  To the contrary, Edward Wasserman, dean of Grad School of Journalism at Berkeley: accelerating towards more sponsored content will only deepen confusion and intensify mistrust.  Back to Nauru: Publishers are getting rid of their most valuable resource, one boatload of shit at a time.

Panel 2: Consumer Recognition and Understanding of  Native Advertisements

Moderator: Michael Ostheimer, Staff Attorney, Division of Advertising Practices, FTC


Chris Jay Hoofnagle, Lecturer in Residence and Director of Information Privacy Programs, Berkeley Law & Technology Center

We’re testing deceptiveness.  Advertorials can be understood as deception by omission/manipulation of schema.  Do native ads cause people to be confused about source of information?  Is lack of disclosure misleading? 

Methods: online survey, inherently not random.  Extrapolation is an issue.  Studying targeted advertising. Bought lists of consumers. One tranche: consumers vulnerable because of situational factors. Another: consumers who had some underlying condition. Another: subscribers to financial journals.  A control group of randomly selected internet users.

We put a native ad in context of a health blog.  Disclosure “sponsored report.”  Not clear who the speaker is. Manipulated background; lacked indicia of being health professional. Asked respondents was the material on diet pills written by journalists/editors, someone else, or didn’t know? 27% said journalists/editors, 43% someone else, 29% didn’t know; no significant differences between groups.

David J. Franklyn, Professor, Director of the McCarthy Institute for IP and Technology Law, University of San Francisco School of Law

Studied what people understand about paid/unpaid ads. Asked questions, showed screenshots of actual search results. Over 10,000 people in several countries. People often skip over labels without even noticing them; a majority didn’t know what “sponsored” meant.  Even when there’s a pop up. The notion that everyone knows, from this morning, is not true—we don’t have homogeneous consumers in terms of what they know or what they want. 60-66% people couldn’t identify paid and unpaid areas of the search results. People remember labels that have never been there. Highly conditioned to see what they’ve been conditioned to see through graphic context. Context matters more than labels. When it looks like a story, people think it’s a story. Context was a different matter online than offline, and it’s different mobile v. laptop.

1/3 of people say they don’t care; 1/3 say they’d click on something more readily if it was an ad. Protecting the consumer from what? If they want to be entertained by a paid placement and don’t care about differentiating—we found real differences in consumer preferences.

We also tested disclosures/disclaimers. They’re highly context specific and wording dependent.  As a general matter, initial attention is higher to labels at the top and left.  We continue to find consistently, in Europe and the US, deep confusion about paid and unpaid.

Jamie Cole, Creative Director, Red Barn Media Group

We do branded magazines for brands. Leverage content out to social media channels. Product is sometimes presented as solving problems but only when the brand owner wants it.  Magazine is presented as a benefit of being a customer, not a sales or promotion tool. Paper on audience reactions to brand journalism: looked at print journals only.  Looked at product involvement to make sure our variables weren’t affected by previous product involvement.  Four groups presented with four articles. Commercial frame with corporate source quoted, corporate name on magazine all the way to noncommercial. Least commercial frame with peer source quoted rated as the most credible; credibility decreased as commerciality increased. Product involvement was the biggest driving factor for any variable. Relevance: difficulty of recognition was one of the biggest issues in our research. Major limitation—how to make the cues clear enough so that participants understand commercial v. editorial. Only 2/3 recognized visual cues for commerciality despite lab setting and clear instruction. More were able to recognize corporate v. peer source but not near 100%. Likely that consumers can’t do it. Unless explicitly made aware of subtleties, weren’t aware of them at all.

Jeff Johnson, Principal Consultant, UI Wizards

Human visual perception and ad-spotting. High-res only in center 1% of visual field; perception is active, goal directed and attention limited; color discrimination is limited; visual hierarchy indicates connection; common vision problems can affect ad spotting.

Everywhere but center, eye has very low resolution: you are legally blind in the periphery.  Hold out your hand and hold up your thumb: that’s the area where you have high resolution.  At periphery, you perceive 3 dots per foot (compared to 300 dpi at center).  Our eyes jump; while they move, they see nothing; when they stop, they see mainly one word and nothing else.

Strongly goal oriented. Where our eyes move is strongly determined by what we’re trying to do. Things unrelated to goal may not be noticed, like labels on ads.

Optimized to see contrasts: edges and changes, not absolute levels. Content designers shouldn’t rely on color/shade. Use redundantly with other cues. 

Visual hierarchy segments page into different parts. Stronger hierarchy = easier to understand what goes with what. “Sponsored links” can look like a peer item and not an overarching label.

Many common vision problems hinder ability to spot ad.  Looking at mobile outside or high sensitivity to glare, ad markers can disappeared.  Buzzfeed: marks ads with color background/gray labels—but if we have yellowing in lens due to lifetime of exposure to UV (over 45) that may be very hard to see.

Dan Greenberg, Chief Executive Officer, Sharethrough, Co-Chair of Interactive Advertising Bureau’s Native Advertising Taskforce

Mostly people aren’t trying to trick you. Macro shift from obnoxious interruption to meaningful content.  (Can’t you be tricked with meaningful content? I thought that was really the point of the disclosure that it’s an ad.) Sponsored stories on FB, promoted Tweets, and everyone else is catching on.  Our ads always say “advertisement,” “sponsored,” or “promoted,” but we’re not wedded to those words. If research says more is needed we’ll find a way to get that into the placements.

Preliminary results from research: does language used in disclosure have an impact on whether or not consumer perceives a story as being paid for? Visuals and context will matter too but we looked at language. Preliminary data: statistical significance in the differences. One case showed a generalized website, some with no ad.  Even if nothing is paid for on the page, a significant percentage of users will say yes to “is there any item on this page that was paid for by a brand?” Sad state of editorial: people would say that a story about Miley Cyrus was “paid for.” People may not know what a brand was. 

Tested mobile and desktop.  Context has a major impact on perception.  Didn’t test demographics.  Can’t decouple from this data whether they didn’t notice something or whether they didn’t notice it was an ad.  Even when we tested “this is an ad paid for by a brand,” we only got up to 70% recognition that it was paid content—but other terms did a lot worse.  Even with just language there are a lot of questions about where to put it.

Michelle De Mooy, Senior Associate, National Priorities, Consumer Action 

Non-English speakers are rapidly growing US audience and especially vulnerable to things like ID theft. Trust is vital for consumers—source really matters.

Chris Pedigo, Vice President, Government Affairs, Online Publishers Association

Represent 60+ premium publishers, visited by 100% of online population each month. By the end of this year, 90% will offer some form of native advertising.  70% heard no complaints about their native advertising; 20% only had a few.  Our member companies go to great lengths to label/differentiate.  They look at native ads as another way to provide value to the consumer. They know the audience and work with advertiser to produce content that’s appealing.

Moderator: do consumers perceive ads differently?

Franklyn: there are pluralities. Some consumers say it doesn’t matter to them to know more about whether something is paid/unpaid in terms of trust, clicks. 40% say they want more clear and conspicuous differentiation and that they’d click on paid content less, or go back to it less if they knew the difference.  One takeaway: we now have immersed ourselves so much in this commercial world that many consumers don’t care and enjoy it. They (not all) enjoy the hyperstimulation of ads, want to sift through it.

Cole: we did see increased credibility from noncommerciality, but previous experience with brand influences that.  Atlantic/Scientology: credibility would be affected by previous notions about Scientology, and that would be important regardless of visual cues. Perceptions about brand are longlasting and stable; they don’t change much.  From where is consumer drawing this idea of credibility? Is it from the info itself—well-done, helpful? Is it from previous engagement of brand? Is it from mere appearance/look and feel? Is it from credibility of material around the content?

De Mooy: we don’t often talk about the content that’s missing. WebMD has traditionally been unbiased and has started to use native ads. In that case, along with other financial/health, that missing info has real costs.

Moderator: are there reasons to think some ways are more/less effective in distinguishing?

Greenberg: yes, it’s incredibly important to figure out which way to do it.  On FB, people know enough to be annoyed by Sponsored Stories.  Instagram—promoted photos.  Comment threads are all about the fact that these are ads (though many probably saw and didn’t react/didn’t know) but context matters.  Signals used on one site aren’t used on others.

Johnson: Strong visual hierarchies, like container widgets.  You have to show someone that there’s a scope in which the stuff inside is sponsored. Strong v. weak visual hierarchy as example.

Moderator: when an ad is designed to look editorial, are there reasons to believe even clear disclaimers won’t work?

Hoofnagle: even a disclosure in the title left 27% confused in our study. There are underlying issues: 27% is a sufficient number to be considered a reasonable consumer. Are these likely to mislead a reasonable consumer to her detriment? There are large gulfs between how publishers and advertisers are talking about consumers and how consumers perceive these disclosures. We heard publishers say their readers are smart. This is no doubt true but in some sense irrelevant because the question is whether some percentage of reasonable people are confused; even smart people may come to different conclusions about what “sponsored by” means. When he thinks of sponsorship, he thinks of PBS. He does not think that BP told the McNeil-Lehrer hour what stories to run; rather he thinks BP provided underwriting. But this morning he learned that advertisers say that stories should be compatible with their products. That’s a completely opposite mental model.  Finds that totally confusing—if the direction of the story comes from advertiser to publisher, rather than the other way around, that doesn’t seem like “sponsorship.”

Franklyn: there’s been an inversion in the relation between content and advertising; what people might have thought it meant before no longer does. We recently tested pop up disclosures by search engines. Roughly 44% of people thought “sponsored” made them more confused about relationship between paid and unpaid content on the page.

Pedigo: not an attempt to deceive consumers. Our audiences are engaged and will speak up about a change in font. In our survey, 71% of members haven’t heard any complaints because they’re doing a lot around transparency and because it’s attractive content.  (I can’t tell how deliberately he’s not listening to the evidence being presented.)

De Mooy: how do you complain about that? To whom?

Pedigo: Oh, they find a way.

De Mooy: Many lower-income people use low-bandwidth, small devices—perception difficulties are already inherent, then add in language difficulties—many disclosures will be useless.

Franklyn: the winner in terms of clarity was “commercial ads”; others grouped.  Had to be sufficiently large lettering, in the right place. Didn’t win by a ton—13% compared to 6%.  Native ads have come on a platform of monetization of search from ten blue links to up to 70-80% of the page ads, some of which say “ads” and some of which say “sponsored.” People have to find what they want in that soup. When you ask them to disaggregate which signals tell them what’s going on, it’s difficult because they’re already conditioned.

Trust isn’t that important practically/in a business way. What’s important is migration of consumers with the brand through new iterations—you can make a very successful business with very partial trust, and that truth needs to be told. You can’t rely on the promise we kept hearing this morning that “we have to be trustworthy or it would be bad for business.”

Johnson: In studies, many people just didn’t see the labels. Everyone here should sit through a usability study of people asked to do a specific task online. You will be amazed at what they don’t see when their brain is engaged in that task. That’s where a lot of the study noise comes from: people don’t see 90% of what’s on the pages they visit/click on.

Greenberg: historically advertising proclaims itself as advertising through interruption.  No choice but to realize it—if there’s a homepage takeover, or a preroll/interstitial (especially on mobile).  What happens when ads aren’t as obnoxious?

De Mooy: language is practically worthless; not even worth FTC’s time. It’s design, context, and straightforwardness of commercial advertising as interruption. That’s the way to move forward.

We’re used to advertisers paying for the shell; people can understand that they paid for the presentation of the content, but not for the subject matter/the content itself.

Franklyn: we will test if consumers understand/care about who wrote the content.  “Sponsored by,” he thinks, won’t materially increase that sort of awareness.

(someone) no silver bullet for different platforms, kinds of audiences.  Terms work differently for 16-year-old girl than for Home & Garden reader; publishers know their audiences, and can work with advertisers.

Hoofnagle: Many companies that used “sponsored by” used a grey color and a smaller font; this is important. Also think about how people think about “partnership.”  HuffPo Partner Studio: the claim is this clearly discloses to consumers that this is an ad. I wouldn’t think that at all!  Lawyers don’t think that. Partners aren’t at arm’s length, but we had one panelist this morning say “we worked with a partner at arm’s length.” Also, intent doesn’t matter. FTC doesn’t have to prove intent to deceive the public. So you may not be setting out to deceive; the question is whether you are misleading consumers to their detriment.

Franklyn: there is detriment to some people. Who will you choose to protect?

De Mooy: sometimes the issue is what content is missing.  Pharmacos pay for pills, but no one pays to educate consumers about holistic methods.

Q: Use of labels like “what’s this?” to ID native advertising.  Or ads only ID’d by Ad Choices logo—does that work?

Franklyn: only 11% are likely to roll over icon for explanation, and of those 44% were more confused when they did.  Icon rollover is a low baseline for getting attention.

Greenberg: think of conversion funnels. Traditionally 100% of people who saw an interruptive ad realized it because there was no conversion funnel. In a FB feed, you see the ad, then your brain recognizes the story before you recognize it’s an ad; disclosure is maybe needed before I decide to click.  Only 50% read the story they clicked on, if that.  Now you’re down to a much smaller percentage who sees the ad, much less realizes it’s an ad.

Johnson: the icon was never intended to convey meaning to those who didn’t know what it meant. Its intent was to remind you of the function that you already know about—like printing, deleting, whatever; it’s extremely difficult for any graphic artist, no matter how talented, to create an icon that conveys meaning without prior knowledge.

Q: what design techniques might make it more or less clear that something is an ad?

Johnson: strong visual hierarchy, boldness—all they can do is increase probabilities; they can’t guarantee anything. Eyes move semi-randomly according to people’s goals.  Movement will move eyes in the direction of movement (periphery needs to check for danger).

Cole: context—if you’re trying to make content look as much like the content around it, the less you disclose the more effective it will be.

De Mooy: doesn’t that mean you’ve effectively tricked people?

Cole: yep.

Franklyn: if you’re talking about internet search, we found that chopping up the page in a more clear way—ads are only going to be on the right side, algorithmic results only on the left—if you could have architecturally mandated segmentation, then people can learn what things are. How that applies to native advertising is unclear because it’s deliberately mashed together. Architecture is not a solution for that fully paid page.  Pop up warning you’re on a paid page? If the goal is clear differentiation/consumer understanding, that’s very hard because the market has overwhelmingly blurred lines and consumers have accepted it largely because search is free and it’s stimulating. 3- or 4- or 5-sided market; consumer acquiesces in use of personal info for targeted advertising, and as long as that bargain is on it will be very hard to regulate.

De Mooy: some analogy to Do Not Track: bring it out of the shadows.

Pedigo: consumers do have choice. If they feel they’re being duped by native ads, there are a million other websites for any other content. Our members are very sensitive to this. If they lose consumer trust, they lose out completely to Joe Blow blogger down the street.  (I’m with Franklyn on this one. Moralistic and convenient to say, hard to believe.)

Hoofnagle: Homo economicus does not surf the internet. Real people don’t have that perfect option weighing available to them.  There’s not a real market where they can understand the price and incorporate that into their decisions—cf. FTC guidelines on use of the word “free,” which discuss its powerful psychological effects on the listener.

Q: will certain populations have more trouble recognizing native ads?

Answers: seniors; lower socioeconomic status groups; varies based on race.

De Mooy: poorer people have smaller devices, slower download times; unfamiliar language. Spanish language sites are often aggregators of translated information; very unclear where information comes from. Hard to tell even how to figure out where it came from. We know that people in underserved communities are at more financial risk of fraud online. They deserve special consideration in regulation. Financial advice and health impact are special categories.

No comments: