0 Survey Looks for Hostility to Behavioral Targeting, and Finds It

Summary: a new survey found that most Americans oppose behavior-based Web targeting. The authors clearly had an agenda, but the industry still needs to present its side of the story.

A recent survey conducted by professors by UC/Berkeley and University of Pennsylvania concluded (to quote its full title) that “Contrary to what marketers say, Americans Reject Tailored Advertising and Three Activities That Enable It.” Is it just me, or do I detect a bit of hostility?

The headlined finding of the survey was that 66% of respondents said they did not want Web sites to show ads tailored to their interests. As eMarketer pointed out in its article on the study, this conflicts with other surveys have found consumers receptive to targeted Web ads.

The Berkeley/UPenn study claims it is more accurate because it is the first nationally representative telephone (rather than Internet-based) survey on the topic. I doubt that really had much impact on the results. As a detailed critique by the business-friendly Progress & Freedom Foundation points out, the answers were more likely influenced by the structure of the poll itself, which asked a variety of somewhat tendentious questions leading up to the final answers.

My own critique is that the questions all asked whether people wanted to see tailored ads, not whether they preferred tailored ads vs. non-tailored ads. People may have interpreted the unstated alternative as no advertising at all. In that case, their rejection had less to do with tailoring in particular than with the near-universal dislike of advertising in general.

Still, the answers I found most interesting have received relatively little attention. This was a set of three questions that found:

- 67% of Internet users agree they have “lost all control over how personal information is collected and used by companies”, but

- 58% feel “most businesses handle the personal information they collect about consumers in a proper and confidential way” and

- 54% agree that “existing laws and organizational practices provide a reasonable level of protection for consumer privacy today”.

In other words, people have finally accepted Scott McNealy’s famous advice from 1999: “You have zero privacy anyway. Get over it”.

I’d like to end the story there, if only for artistic reasons. But the survey also asked several questions about consumers’ understanding of current privacy laws, which basically found that most people think they have more protection than they really do. Another series of questions found support for laws giving consumers rights to insist that companies delete their information.

I don’t take the results too seriously because the questions didn’t indicate these would be new laws or balance the laws against reduced free content or the cost of more regulation. But they do suggest that people might support stronger regulation if they understood how poorly they are now protected. So there continues to be a real need for marketers to both do a good job of protecting consumer privacy and of educating the public and legislators about the benefits provided by easy access to consumer information.

0 More Surveys Agree: Web and Non-Web Data Must Be Integrated

It’s not that I’m obsessive, but just to gnaw a bit more on last week’s bone about the coming unification of Web and other marketing data...

- a recent survey sponsored by enterprise marketing automation vendor Unica found that “integration with other marketing solutions” was the most commonly cited web analytics challenge (46%).

This was followed by “verifying accuracy of data (inflation/deflation)” (41%) and “not comprehensive/missing types of data” (32%). It’s interesting that the question was about Web analytics in particular – even analyzing Web results by itself requires non-Web data.

- another survey by another marketing automation vendor, Alterian, found the most-commonly cited top obstacle in online marketing (25% of respondents) was “integration of online with database marketing and offline channels”.

Now, this isn’t exactly the same answer as the Unica survey, since the question isn’t limited to data integration. But Alterian also reported that “lack of ability to assess or manage internal infrastructure and culture challenges (25%) and the integration of all the technology to power the cycle (20%) were identified as the biggest factors in implementing the customer engagement cycle.” So clearly data and system integration are indeed primary concerns for multi-channel marketing.

For still more on this topic, see today’s post on the MPM Toolkit blog, describing a very detailed report in eMarketer about Online Brand Measurement. This provides still more evidence for the need for integration of Web and non-Web data, in addition to other issues.

Case closed.

0 Web Analytics Is Dead. So Is Customer Centricity. I Need a Drink.

Summary: Web analytics is merging into the broad world of marketing measurement across all media, which itself is shifting focus from tracking individuals to understanding group behavior. Although Web analytics and marketing automation vendors are currently wrestling over who will house customer data, both are likely to lose custody to enterprises who want to control their data for themselves.

Even as analysts are still sorting through the implications of last week’s acquisition of Omniture by Adobe, the industry saw two additional important announcements this week: Omniture combining its data with comScore to help measure Web advertising audiences, and Nielsen working with Facebook to poll consumers on advertising impact.

Both announcements share several interesting features: they don't rely on traditional Web analytics (tracking page views); they involve vendors who report data from consumer panels; and they relate to measuring advertising measurement. Maybe they coincided simply because the big Advertising Week conference is now under way. But I think they, along with the Omniture/Adobe deal itself, hint at something more profound: the end of Web analytics as we know it.

Ok, that may not be as earth-shattering as the end of several other things you might imagine. But many marketers are just getting their arms around traditional Web analytics. So it’s worth warning them that things are about to change again.

Smarter Content on the Way

Let’s start with Omniture/Adobe. After thinking about it for a week (and hearing what the participants said), I think the main purpose of the deal was to let Adobe build content that was more intelligent in two ways:

- First, the content will be inherently “instrumented” to report how, when, where and which people are consuming it, using techniques go beyond traditional Web analytics methods (server logs, Javascript tags and cookies). This is needed because content increasingly exists outside of plain vanilla Web pages that can be tracked with conventional techniques. Problems include Flash, video, audio and other non-HTML Web content; venues like mobile, digital video recorders and interactive TV; widgets that migrate through social networks; and plain old cookie deletion. Web analytics vendors are striving to extend their technologies to capture these, but at some point you have to look for a different basic model. Content that can itself “phone home” rather than relying on the carrier medium could be the long-term answer.

- Second, content will be self-optimizing. This involves built-in tests and, perhaps, a sort of swarm intelligence where subsequent views are actually modified based on previous results reported by the content itself. Imagine a widget with a built-in A/B headline test: every time someone accesses it, the widget offers one headline or the other, and reports the result to a central server. (This could be done without transmitting personally identifiable information, so privacy issues are minimal.) Once a pattern emerges, the server could instruct the distributed widgets to only display the winning headline, or, better still, to start a new test. Even without a central server, each copy of the widget could still run its own test and, assuming it’s accessed enough times, adjust all by itself.

Who Owns The Data?

So far so good, and I think that’s plenty of reason to justify spending $1.8 billion for Omniture. But I also think Adobe was interested in the fact that Omniture controls so much of its clients’ data.

There is a battle brewing over that issue: like most Web analytics vendors, Omniture stores Web traffic data on its own servers and sells clients the ability to access that data. That didn’t raise any particular business issues when Web data was viewed largely in isolation. But as the Web takes an increasingly central role in customer contacts, marketers need to merge that Web data with their other customer information. Indeed, if you want to use that data to help guide customer interactions across channels, the data must be not just centralized but also updated in real-time. That means marketers either give all their non-Web data to their Web analytics vendor, or directly capture Web analytics data on their own servers.

The Web analytics vendors see this future and recognize that they’ll be more important to their clients, and thus able to charge higher fees, if they hold everything. Marketing automation vendors and in-house IT groups see the same future and recognize the threat to their own positions. Thus, the business-oriented marketing automation vendors (i.e., demand generation vendors: Eloqua, Silverpop, Marketo, etc.) already capture Web behavior in their own systems and integrate it directly with customer management. The big consumer-oriented marketing automation vendors (SAS, Teradata, Unica) also offer Web analytics, although perhaps less tightly integrated.

The problem here is that the Web analytics vendors and demand generation vendors are largely SaaS systems, so both hold the integrated customer data outside of the client’s own data center. It’s not clear that clients – especially big enterprise clients – will continue to accept this. The consumer-oriented marketing automation vendors already largely support on-premise configurations, so they may be the real winners in this battle. Yet bear in mind that looming behind the marketing automation vendors are the enterprise CRM vendors, who also offer largely on-premise installations. They may eventually gobble up the marketing automation business and the Web analytics data along with it. In that case, Web analytics vendors who hope to become rich stewards of centralized customer databases will be very disappointed.

Customer-Centricity Is Obsolete

I also think there may be one still deeper trend at play here, although this is more speculative. Let’s call it a shift from customer-centric to community-centric marketing. Since customer centricity has been the ultimate goal of marketers, or at least marketing gurus, for several decades, I expect some skepticism.

But think of it this way: marketing has always been about deploying and propagating messages to consumers. In recent years, we’ve striven and become technically more able to target those messages directly at individuals. Yet messages were never really limited to one person. Even if they were delivered privately, they could be shared directly through conversation, physical and electronic pass-along, and indirectly as consumers discussed their experiences with the company in general. Thus, there has always been a community component to marketing campaigns. In cases such as word of mouth programs, this was even the primary objective.

Today, of course, that sort of sharing has become increasingly important for every marketing project. Thus marketers have more need to track the secondary impact of their messages on the larger community. Happily, they also have more technology to do the tracking.

Here’s what’s interesting, though. Marketers will never be able to trace the exact path of each message from one person to another. And even if the data were available, there were no privacy constraints and they could handle the volume, marketers would still face the insurmountable challenge of that multiple messages influence final behavior. That is, they could never meaningfully say that one particular message was the single reason a customer did something. All they can ever do is to look at the many different messages a customer probably received and compare these with actual behavior. If the customer did what you wanted, the messages somehow worked.

If this sounds familiar, it should: it’s the classic problem faced by brand marketers in measuring the value of their investments. Their solution has always been to measure intermediate variables such as consumer attitudes, and to measure these through samples rather than by polling everyone in the market.

This brings us full circle to the panel-based attitudinal research at the core of the Omniture/comScore and Nielsen/Facebook deals. Once you recognize that what’s most important is the broad community impact of your marketing efforts, you fall back on those types of measures rather than attempting the impossible, and impossibly expensive, task of tracing the exact path followed by each individual. In other words, what we’re seeing here is not some Mad Men-style reversion to obsolete brand marketing behaviors, but a recognition that modern marketing is community-driven, so its measurements must be as well.

Now, if I can just find a reason to bring back the three-martini lunch….

0 RightNow Adds Social Community Capabilities (But Don't Expect Support Costs to Fall as a Result)

Summary: RightNow has extended its social media footprint by purchasing HiveLive, which lets companies build public and private communities. It also released a benchmark survey showing that online channels (email, chat, Web self-service) don't do much to reduce customer service telephone calls.

In keeping with my recent posts about broader utilization of social media, I had a chat earlier this week with on-demand CRM vendor RightNow , who updated me on their recent purchase of HiveLive. HiveLive provides a social community platform, which means it lets companies build their own discussion groups, forums and such. HiveLive has many features to support business communities, both in terms of engaging with customers over issues such as product features and bugs, and in terms of building internal communities such as project teams.

HiveLive fits with RightNow’s vision of giving customers a seamless flow between community applications and a company’s traditional service systems. For example, if a question posted to a forum goes unanswered for a specified time, it can be escalated into a service system as a case to be handled by the company’s support group. If I understood correctly, RightNow and HiveLive can do this already.

We discussed deeper sharing, such as having answers developed in a public forum become part of a company’s internal customer service knowledgebase. That’s something RightNow may add in the future.

Our discussion veered onto other topics, and in particular how seriously companies really take the goal of improving the customer experience. RightNow shared a copy of its recent RightNow Multi-Channel Contact Center Benchmark Report, which was interesting in its own right.

One tidbit I found particularly intriguing was how few telephone service calls are “deflected” into email, chat and Web self-service channels. In the survey, most companies reported that fewer than 10% of customers in those channels would otherwise have made a phone call.

You could see this as bad news for the theory that having alternate channels available will reduce the need for call center agents. Or you could consider it good news that customers have more choices to pick the interaction method they find most congenial.

Another interesting item was that 55% of companies have some mechanism to gather feedback from customers, but just 10% of those use an IVR survey, which I personally consider the most effective way to gain broad participation. Again, you can treat this as good news (at least half the companies are trying) or bad news (just 5% of the total are doing it effectively). Either way, it’s food for thought.

0 Acxiom Uses Social Media Data to Segment Email Lists

Summary: Acxiom's new social media marketing tool gathers public data about social media links and uses it to segment email lists. It's a different, and arguably more practical, approach to helping marketers take advantage of social media.

Acxiom last week released a new “social media marketing” solution called Relevance-X Social.

The press release is frustratingly vague (“With the ability to engage socially active customers and prospects in their preferred networks, marketers can link that knowledge to relevant communications that ignite conversations on behalf of the brand.”) But, on talking to the company, it turns out there is a pretty interesting product here. (Disclosure: I am a consultant to Acxiom, although I had nothing to do with this product).

What Acxiom has done – and this is so Acxiom – is to ignore the content posted in people’s social media comments or profiles, and just capture the “hard” information about links between people and membership in groups. Apparently (and I’m taking Acxiom’s word on this), this data is publicly available from most social networks (Twitter, Facebook, MySpace, LinkedIn, Plaxo and some more specialized ones) once you know someone’s email address. So Acxiom has taken its own database of more than 500 million email addresses and found the connections for each.

Relevance-X then accepts a marketer’s own email list – presumably its customers or prospects – and returns information from its own database about the matching names. This avoids at least some privacy and spam issues, since marketers are only given information about people they already have some type of relationship with.

The main application of this information is sending targeted emails. Thus, a bank might send one message to customers who belonged to a financial planning group, and a different message to customers who don’t. Other segmentation might be based on the total number of connections, membership in the company's own fan group, or information the company already knows from other sources.

The key point here is that Acxiom is using social media to execute traditional database marketing. This is quite different from most social media marketing products, which boil down to monitoring for posts on specified topics, responding to individuals, or to publishing messages to groups through the network itself. In a way, it seems rather old-fashioned to use social media data as a basis for outbound marketing. But for marketers struggling to find a practical use for social media, it's better than many alternatives.

(As Ed Park points out in a comment below, other vendors including RapLeaf and Unbound Technology also build similar databases by capturing social media links.)

Relevance-X includes two other components. One is the ability to tag the content it publishes – such as links within emails or messages posted to group pages – so marketers can track response. This is done with standard page tags and browser cookies, so what’s important here is not the technology but the ability to measure results. Again, this is something that traditional database marketers consider essential – and that other social media products sometimes struggle to accomplish.

The other component is a separate social media monitoring service that tracks keyword mentions, sentiments and trends, but on an aggregate basis rather than by tracking individuals. Acxiom is using a third party product for this. The goal is to supplement the direct response tracking with a more general measure of marketing program impacts.

Pricing for Relevance-x Social is based on the number of relationships (typically email addresses) researched and on the number, size and complexity of the campaigns being managed. It can be purchased on a campaign-by-campaign basis or annual subscription. Pricing for a basic campaign could start at around $25,000.