There is site engagement, which Nielsen measures as the average amount of time visitors spend on a website.  At the site level, many would argue that engagement is a reflection of time AND activity – how many clicks were there (click depth), did the visitor post a comment, how often does a user come back, or did they perform a desired action. Simply measuring overall time spent can be a misnomer considering poker sites may keep users engaged for hours at a time with virtually no page clicks.  Eric Peterson has done an excellent job of defining site engagement using six normalized metrics that combine all aspects of website interaction, which can be found here.  He defines engagement as: an estimate of the degree and depth of visitor interaction on the site against a clearly defined set of goals.

Another aspect of engagement is campaign engagement.  As consumer apathy towards banner ads increases, advertisers are increasingly turning to interactive ad units consisting of games, original content, and social features, all of which require new measures – such as how many pass-a-longs, or interactions, did the campaign generate.  Facebook has launched a product called engagement ads and companies like VideoEgg are launching pay per engagement based buys.  Paidcontent.org did a nice piece on some of these emerging models here.  So the ‘it’ in this case is the ad unit and not the website.  Depending on the advertiser’s objective -  sales, clicks, or simply share of mind – applying these various metrics to achieve an overall engagement score is possible for both website and campaign engagement.  As a recent MediaPost article on engagement stated: these metrics ultimately measure consideration and intent, which translate to sales and market share, but they should be measured across all brand related campaigns.

But what about the content or the brand itself IN the content?  If the content isn’t interesting, the website surely won’t be engaging.  And if the website isn’t engaging, I seriously doubt whether engagement based advertisers will want to spend their budgets there.  So how do we define if a piece of content is engaging – or rather which content is engaging?  One of the questions I always ask our clients is how well their latest feature piece is doing.   The answer is always in page view stats and the comparison is made against prior feature articles.  The problem with making this comparison is that it is being made in a vacuum.  If Rolling Stones does a lifetime piece on Stevie Wonder and their average engagement level is 121 seconds and 6 posts, is that good?  I can tell you it is because across the 45 music sites where Vertical Acuity directly measures Stevie Wonder content, he gets an average of 72 seconds of engagement which is much higher than the average R&B artist’s 32 seconds (ironically at the time of this post the number one R&B artist was Bobby Brown with 90 seconds – I wonder what he did this time..).   This is shown in the screenshot below.  Shouldn’t that content be worth more to an advertiser knowing that?

Stevie Wonder Average Engagement

As a publisher, wouldn’t you want to feature content about subjects that people find more engaging, and as an advertiser wouldn’t you want your campaign to appear next to subjects you know will get your brand two to three times the average exposure if engagement really is becoming the measure of choice?  At Vertial Acuity, we believe the cornerstone of measuring engagement comes from the content, which drives the website, which drives the advertising.  Our goal is to help our publishers utilize subject level engagement metrics to improve both their content targeting and their advertising revenue.  What are your thoughts on engagement?

Traditional website evaluation methodologies use Web analytics to determine what is “working” and then use a Content Management system to manually identify, tag, and deliver the content websites think consumers want. Tagging content is a manual process and fraught with user error, which is why content is often miscategorized. New types of analytics and tagging technologies are required to dramatically improve this process by analyzing a websites content at the most granular level, the subject level, and providing those websites with new views of how their content performs in comparison with their industry.

For example, we determined for one of our clients in the radio business that three photos of hot celebrities posted and linked in their morning show blog were representing 45% of their page views during a 3 day period in comparison to their other radio station websites – this content could have been immediately leveraged across their other properties with similar genre and customer focus to improve page view performance. Website owners don’t know they are missing valuable content or products if they have no way to measure it in comparison to a broader, more standardized set of measures.

So while our semantic analytics continue to help our customers determine hot content (subjects) they should be featuring on their website, we wanted to take the next step and help them automate the process of acquiring, organizing, and presenting their best content to their visitors. SmartThreads has been in beta testing with some our premier customers, and we continue to improve on our initial results (measured by Page Views per Visit and Click Through Rates) by incorporating more and more of our subject level measures into the content targeting system.

We also moved office locations during the month of April and are really enjoying the new space and 18th floor patio overlooking Hotlanta. Luckily for us our neighbors happen to be in the wine business – nothing like a wine tasting with a view. We titled this photo taken from our patio, “Atlanta in a glass”.

Atlanta in a Glass

We look forward to more frequent updates as our targeting product goes into full scale Beta.

From Vertical Acuity’s beta music measurement network, we can see that Notorious B.I.G. (in the following screen shot) started to see a massive increase in traffic to his content starting around January 3^rd which peaked around January 7^th. Content articles analyzed included everything ranging from movie related articles, to track postings, to photo pages. So what do these numbers represent?

Notorious B.I.G. Traffic Summary

For the week of January 1, Biggy experienced a 1400% increase in web traffic to his content with the typical consumer spending an average of 28 seconds reviewing Biggy related content articles. While the majority of his page view gains occurred from website home page postings, music related pages containing his album and track information came in second. His top 10 viewer cities are listed above with Atlanta receiving 19,710 page views of his total 857,251 during the period – a fairly even geographic distribution nationally. Biggy’s velocity is 1.64 page views per session, which means the average user will read 1.64 pages of Biggy content during a visit. This is on par with the industry average of 1.67 page views per visit for the top ranking 100 artists (ranked by total page views). We also noticed that some of the publishers within our beta measurement network took advantage of this trend by featuring Notorious content on their home pages, which significantly boosted page views for their sites during the period.

We also noticed around January 1, 2009 that Fox Studios began running major campaigns across music and entertainment sites for the upcoming movie of Notorious B.I.G.’s life. While these campaigns and other PR related activities can help explain the rise in consumer interest, previously there has been no way to quantify increases in web traffic to a particular artist like Notorious. To do this would require assembling web analytics data for 100’s of different URL’s representing Notorious content across dozens of music sites.

So the first question we might ask is, isn’t this the same information we can get from Google Trends or from a company like Hitwise, or from one of the companies that monitors consumer buzz? While we will dedicate a future post to looking at the online measurement landscape and different companies represented in that landscape, the short answer is no – aggregated traffic data across major points of consumption (music sites) is not available by subject. Let’s compare this with information we can get from Google Trends:

Google Trends for Notorious B.I.G.

While Google trends does show a significant increase in search volume during the same period, this only represents people who are trying to find Biggy content – what happens after they leave Google search, or the people that go directly to music sites versus using a search engine? Where is he most effective? If you were Fox, what markets would you target and how well did his campaigns do? The differences in the top 10 cities between searches and actual consumer views of Biggy content highlight that the geographies where people search for subjects are not necessarily the ones where people read the most about a subject.

This also highlights the differences in using subject level market intelligence data as a source for geo targeting versus search engine data. One supports onsite display and banner advertising, the other search engine marketing. Considering less than 25% of internet users utilize search engines to locate entertainment subjects (according to Hitwise data on search engine referrals by industry), it is critical to mine the wealth of information that comes from the other 75% to improve contextual targeting with behavioral trend data.

In our next post I will discuss where we feel Vertical Acuity fits in the measurement ecosystem, and compare and contrast the differences between Web Analytics, Market Intelligence, Consumer Sentiment monitoring, and Ad Serving technologies.

Vertical Acuity defines Semantic Analytics as, “The analysis of related website content and anonymous user behavior data with respect to any subject appearing within a single webpage, website, or group of related sites”. A subject can be defined as a brand, a product, an artist or album, a travel destination, or even a car model – any individual item that can be tracked on the Internet. By tracked, we mean actual measurement of how many people are reading the subject, where they are coming from, and how much time they spend with that subject across multiple websites, etc. You can think of it as a normalized approach to web analytics, also referred to as the semantic web when looking at the relationships between related subjects. Subjects and subject relationships can be rolled up into categories and sub-categories – the brand Coca-Cola is a soft drink or beverage. The Apple ipod can be categorized under consumer electronics and further under portable music players, and the artist 50 Cent and his albums are categorized within the Hip-Hop genre.

By breaking down a webpage into its primary subjects (what people are researching, reading, listening to and watching) and aggregating web analytics data at the subject level with additional sources of data such as consumer sentiment or demographics, a 360 degree view of a companies products, digital assets, and brands can be created across like sites within an industry category. To create a complete picture of online product and brand performance, six components must be measured:

1. Occurrence – how many times does a product or brand appear and on what types of sites and pages

2. Demographics – who is viewing the product or brand and where are they located (Geographics)

3. Velocity – what is the growth rate of a product or brand being mentioned, as well as consumed (actual views), and on what types of pages

4. Engagement – how many seconds to people remain engaged with a product or brand

5. Reach – how many people is a product or brand reaching during a given period of time

6. Location – where does a product perform the best

Once this information is consolidated, it can be used in a number of ways, including: improved targeting of content and advertising, measuring consumer engagement and advertising effectiveness (think measurement of ad latency), predicting products, brands, artists, etc. that are close to the ‘tipping point’, understanding which search terms are used to find products from both search engine traffic and network traffic – the list goes on.

A good illustration of how using semantic analytics data, in the context of market intelligence, could improve targeting can be illustrated in a recent article on Clickz entitled, “The Trouble with Audience Measurement”. I think the article effectively highlights one of the major problems in the industry by citing an automotive targeting example. The author discusses an example where an automotive company wants to reach women reading about autos who are in the middle of the purchasing cycle, and upon selecting a site to advertise on finds out the sites audience is 60% male after running a Nielsen report. Why does measurement stop at the site level? Because of panel sizes, lack of data, technical approach?

There are a number of reasons online intelligence stops at the site and category level, and I will cover those in future posts, but the key point I want to make with this reference is around targeting automobile models that have a higher percentage of female readers rather than websites. Why can’t we produce information that shows which makes and models have a higher percentage of engaged female readers across automotive sites and then target those subjects? Or maybe only those female readers that are actively engaged with certain makes and models across sites. Why can’t we ‘cherry pick’ those audiences based on subject level data trends versus picking sites based on a 60/40 demographical split? Site Owners should have the ability to slice up their site by key subjects, use the data to find the right audience segments, and stop wasting impressions on a 60% gender breakdown of their sites content. We believe the end result for many websites would be an increase in advertising rates while allowing marketers to reach the exact buyer profile they want to reach.

Future posts will discuss this and other areas in more detail, but our next post will look at an example subject performance profile for the artist Notorious B.I.G and provide some insight into his online performance from our beta measurement network.