Archive for the ‘audience measurement’ Category

Wow: Google sneaking in the backdoor on Digg and others!

A seemingly innocuous post on Google today regarding long forgotten service Measure Map.
Thank you Tech Crunch, this was a phenomenal bit of news that could have been lost in the ether.

However, is it really so innocuous. Let’s see what the service does….in exchange for distributing a Google Analytics pixel on your blog, Google can now offer you stats about your blog interaction. Seems relatively benign, however let’s extrapolate the value and just for fun talk about some of the companies and “spaces” that this can meander Google into.

By allowing users to categorize their blog and merely just by putting the pixel on, Google can aggregate even more signals at a higher frequency about a user interaction with content (in exchange for data reporting).

Possible ancillary / competitive services:

1) Well there is Digg. True it is user that contributes to a portal regarding a news stories worthiness. All Google really needs to do is an opt in service and portal overlay for “find the latest popular stories on Twitter.”

2) There is also Omniture and other web tracking companies (but this one has been happening for some time)

3) What about a BuzzLogic effect whereby Google offers data back to advertisers regarding popular blog posts (on an opt in basis from the blog owner of course)

4) How about Comscore and other publisher-side measurement services (Compete, Alexa) and offering publisher side data on what is transpiring on my blog?

There are many more, however it will be interesting to see the traction this service gets. I intend to review the service and thoroughly review the privacy policy for data usage once it becomes available.

As a note, I am continually impressed with Google’s fastidious detail to aggregating signals.


Adify: Selling at the apex before the flight to quality

News today that ad platform company Adify will be bought by Cox Enterprises in an all cash deal.

Congrats to the Adify team on the deal. They assembled quite a bit of value in a short period of time.

It also seems that there sales comes at the right moment. Vertical network plays work by allowing a blend of brand and performance-based advertising; having the brand though is the key.

For example, “Sustain Lane is a green community/network/marketer, however you see it. They need to sell a certain amount of green advertisers to have their effective CPM be, well, effective.

If an ad recession hits, and the first tell tale signs are “WebMD’s numbers last week, then smaller and newer networks (with less volume and history) will have a tougher time earning brand dollars and continually justifying a platform.

This is not say there is doom at for Adify’s program. By bundling an Adify solution with TV inventory or at least the verticalized cable advertisers, it creates an integrated campaign where Cox can create more value.

So a nice win both ways.

This brings a solid base hit back on Adify’s $27M in funding. There should be an indirect corroloaries reviewed to the Rubicon Project and Pubmatic and other ad service-type companies.

Internet Marketing Observations: YOU OWE US NEILSEN

So for those that follow this blog, we ran a post last month that called out Neilsen and Clickz for publishing a report on paid text links by spend that could not have been true.

The title of the post: Neilsen Online: At what point are you embarrassed?

I happened to click on that story today, because it’s skyrocketing up the hit chart today on this blog.

I went back to the Neilsen data source on Clickz.

And to my shock and astonishment, I found a disclaimer around TableofSix being listed as advertiser #10.

It reads, “*After these data were published, Nielsen Online said it learned that the impression counts for Table for Six were inflated, citing an ad collection issue with its sponsored link impressions on It said the issue has been fixed and the data should be normalized effective the week ending April 20, 2008. “

Could we please get a thank you over at IMO for qa’ing your data Neilsen. Next time, you do the homework.

Google top brand? Sure, but what’s the methodology.

Since my entrance into online advertising some 10 or more years ago, brand metrics have always been focused on reach, uniques and impressions…..and that’s about it. Occassionally there is the media buyer who requests a site-by-site agreement on where they are to run; who knows how they pick which Web sites they run on.

As brand marketers start focusing more attention online, it’s only natural and, in a push for transparency, for advertising sales execs to request: “What are your success metrics?” and specifically how they will be determined.

Online advertising is about accountability and that will not change. You will always be able to track and report on impressions much better than any other medium.

With that, I am wondering what to make of Millard Brown’s announcement of Google as the top brand. I’m not disagreeing that they are, but I want to learn a little bit more about their survey and methodology so as to begin creating better metrics for online advertising.

Taken on the surface, the survey represents a very, very small sampling size. The 100,000-person survey when compared versus online audience is less than .01% of the Internet audience (assuming everyone surveyed has an internet connection).

It’s time to push brand metrics to balance higher production endeavors like NBC-Omnicon. Clients should demand this.

AdRoll: Effort level vs. volatility

Those who read this blog know I am a fan of the vision of AdRoll in terms of solving for the two most important macro issues going forward: Small publishers and branded marketers. I am not affiliated with AdRoll in anyway.

From the press release and TechCrunch coverage it appears they are truly focused on bridging the pain points on both sides.

AdRoll will work because of the power of ability to manage groups. Some of the tough operational things that will face AdRoll as it scales will be: fraud, managing “group participation” rulesets and with that, volatility.

Some of the more global competititive issues facing them will be adserver integration and where they sit in the daisy chain of decisioning. I thought of this while reading the TechCrunch message board interaction between Pubmatic’s and AdRoll’s CEOs. It is interesting to note the language that the Pubmatic CEO used, specifically “we look forward to adroll as another option for publishers.”

At the end of the day, one of the companies has to come first in the daisy chain of decisioning and I would surmise, given that AdRoll wants to cater to brands, that they need to be the first option.

Both Pubmatic and AdRoll give smaller publishers a much better option than a blind network in gaining more for their inventory. The extent they can make this scale across a wide swath of brand marketer dollars will be the key to both systems.

Hopefully neither business model will come down to what VP of Sales joins or the tempermental nature of brand dollars.

Neilsen Online: At what point are you embarrassed?

Actually, I’m not sure if I would be more embarrassed if I were Clickz or Neilsen at this point.

I visited the Clickz Web site this morning and clicked on the story Advertising Placements by Industry and Top Sponsored Links, February 2008.

I actually always get a good chuckle out of this. To be honest, all I did at a former company was make sure to keep purchasing media on Yahoo because it artificially inflated how much money our company was spending to potential advertisers. For some reason, the Neilsen panel skews to Yahoo users.

But, I’ve written about this before.

However, as I was scrolling through the running “commentary” this morning, I affixed my eyes on advertiser #10 in terms of sponsored links: TableForSix.

For those of your not familiar with <a href=””TableForSix, it’s a unique spin of non-direct online dating centered around a dinner party. It’s actually a good niche business. (I actually looked at the Wayback Machine and found out they launched in 1999.)

More importantly and to the point, both Quantcast and Compete have their uniques per month hovering about 45,000.

45,000 uniques.

Neilsen: 900,000,000 impressions

Does that seem plausible to you? No, not that TableForSix is a top 10 advertiser, but 900,000,000 impressions turning into at most 45,000 uniques, maybe 90,000 page views.

Quick data review for effect:

Neilsen: 900M impressions

Compete/Quantcast: @45K uniques



Visits ratio: 2 per unique (Quantcast)

Visits: 90,000

CTR: .1%

Clicks per linked impression spend only: 900,000

Best case scenario: Neilsen is 10x off

Again, I don’t know what I find more incredulous here:

– that Neilsen data is still treasured by marketers, especially brand marketers
– that Neilsen publishs and promotes data that doesn’t even pass a sanity check
– that Clickz promotes Neilsen data without sanity checking the data itself

There has been known to be issues with panel data for ever. However, that does not mean that both the industry and the companies promoting that data should sit idle.

A fair tradeoff here would be transparency in how Neilsen will improve this data over time. That should not be too much to ask.

And please Clickz, don’t prominently promote this data without a review, it does your web site, your brand, and your readers a disservice.

Follow-up to the Dashboard post: What’s your CPM?

There was a comment on the dashboard post regarding brand marketing vs. direct response marketing, that demanded a follow-up post.

Having been working online for more than 10 years here, I can say without a doubt that performance-based marketing, and Google itself, has driven internet marketing and content production more than anything else.

I only need to look to major equity deals like NexTag’s billion dollar private equity deal,, heck even aQuantive-Microsoft as a few examples.

I hate to keep pointing to other people’s blog for fear that this blog lacks value, but I’ll point to expert Andrew Chen’s blog one more time here because he eloquently describes the need for brand advertising to catapult a business model.

The reality with performance-based marketing is that it’s very iterative, competitive and non-service based. I’ll use this example, I am purchasing the keyword “Las Vegas hotels” on the top spot at Google. Let’s assume for example that my creative is a mirror image to the client’s below me. How do I get that spot? Obviously by bidding higher.

The point, small increases in effective CPC or CPM, can power large gains in volume. Let’s use an EDU example, I’m NexTag bidding versus ClassesUSA for the Yahoo Marketplace position. If I bid $.05 more per impression (or even less!) I own vast amounts of volume. Assuming that placement averages 200 leads a day at $60 a lead, that $120,000 in revenue that I’ve been able to earn by bidding just a fraction of a percentage more.

The point, gains made in online marketing are iterative and need to be well coordinated.

In the simplest terms, I need to own the highest RPL or RPA (Revenue per lead or acquisition), own the highest conversion rate, and own the highest Cost per Acquisition vs. Volume metric — or I at least need the aggregate of these to be slightly better than my competitors. That takes technical sophistication (products and integrations), expertise, and competitive strategy. It’s a lot of heavy lifting….daily.

Which brings me back to my point? Say I’m buying inventory on Yahoo Autos! and arbitraging that traffic an effective lead price or cpm arb price, I can bid $2.00 CPM for that traffic and I’m constantly fighting off my competitors at $1.96 and maybe $2.01.

However, I have absolutely no answer for Lexus who comes in there and buys that exact same inventory at $7 – $14CPM (at that’s probably ROS, non-roadblock). How do I possibly manufacture a competitive $7CPM on member value or lead price. I can’t.

90% of marketing dollars–I’ll check that fact–are spent on brand advertising. It’s the reason that Google decided to bite the bullet and partner with Publicis. Their thinking, “Well I can’t come up with a great product yet for all brand advertisers, in fact most of them fear me, at least I can move the needle by partnering with one of them and getting some spend.”

The biggest opportunity online is in successfully scaling a brand marketing solution. A solution that brings those double digit CPM online in earnest and with consistency.

Any “solution” directed at performance marketer needs to provide such a consistent and measureable ROI impact immediately upon integration that increases conversion exponentially and I just don’t see that happening.

Thought I’m not sure if I see someone able to short-circuit the brand/agency sales cycle either, but if the product’s there, then there has to be a good salesman somewhere.

A quick note here: A thanks here for everyone who has taken to reading this blog and providing feedback. Getting thoughts on paper, or screens rather, is great in terms of hearing other people’s opinions on Internet marketing.

Building signals with Twitter (and for Google, with Digg)

Update: More on “Google looking to buy Digg, maybe..

What is Digg to Google? How about inmarket indicators for where display advertising is about to ramp up. Let’s call it Yahoo Buzz for Google?


I stumbled upon (no pun) the Twitter community page on Silicon Alley Insider this morning.

It was immediately then that I figured out how Henry Blodget was expecting to go up against Michael Arrington in tech news (or so I believe). Silicon Alley would use Twitter signals to identify newsworthy companies before TechCrunch (or so if my thesis again).

A common data aggregation and processing point for many companies, like Google, is building signals. In this case, signals are bits, sometimes incomplete, of data that a company coalesce together to create inference or data review. By understanding signal type, characterizing it, and monitoring its frequency, a data system or company–in this case a small one like Silicon Alley–can interpret actions before they happen.

In this case, Silicon Alley could here about Microsoft-Yahoo deal (through evaluating Twitter messages) before multiple sources from TechCrunch. Isn’t this also what Digg is somewhat about?

In a bigger more important, Twitter-type signals could be used to learn about a tsunami as it’s or before it’s happening.

The interpretation of signals in business and through communication on the Web (Twitter, blogs, etc) is surely a burgeoning industry and one that will get more press as computing infrastructure can manage more data in shorter timeframes. The challenge will be in accurately labeling the data in realtime so as to create meaningful interpretations.

More on signals and brand metrics here:

Brand Advertisers, Your Move! What’s valuable inventory?

There have been many stories lately focusing on how to evaluate your inventory if you are an online publisher. Major publishers dropping adnetworks, social media CPM compromising Google’s numbers, announcments by publisher service firms on pricing, etc. etc.

What all these get to is defining what is valuable inventory. Again, ask a performance marketer and they point specifically to “gross profit dollars” in evaluating a placement. Performance marketers are only as good as their latest collection of creative or offers and how is performing versus a segmentation of their inventory (by publisher, by placement, by creative, etc.)

Ask a brand marketer, who is paying 10 – 20x the rate for the same inventory and usually you get things like unduplicated reach and frequency and a soft review of a metric.

Andrew Chen, an expert in the field, has suggested that equation for evaluating your revenue is the following:  (p.s., I encourage you to read and subscribe to Andrew’s blog)

Brand revenue = # campaigns sold * average campaign size * brand CPM
Direct response revenue = (total impressions – brand impressions) * remnant CPM
Total revenue = Brand + remnant revenue

Here’s the post:

I, of course, agree with him, but it also begs the next question:

How are those impressions displayed?

Case 1: A brand advertiser might purchase a 7-day buy on the front page of for example, however there may be 10 other “advertising” placements by on this page, either brand or performance based.

Case B: The same brand advertiser may buy on the “New Cars” section and there may be zero placements from brand or performance marketers on these pages.

Which inventory for a brand advertiser is more valuable? Does it matter?

If there are no standards, what I am doing if I am a publisher focused on ROI is coming up with a way to optimize inventory that:

– creates many many impressions

– optimizes performance marketing on pages that *have* brand advertisers

Does this sound like inventory that a brand advertiser wants?

It’s time for brand advertisers to take accountability and describe either through standards or IO the type of inventory they want and what it means for their brand. The onus is on them. Further, the onus is actually on advertising agencies specficially who control a vast amount of spend for brand-based advertisers.

This should be part of the service that they are peddling to their clients.

More on the value of inventory later this week, including a follow-up to this piece:

Focusing on the supply: Good move Pubmatic

Very nice press release by Pubmatic this morning. The company is focusing it’s business on helping publishers manage their business.

An expression I am fond of using nearly every day, is that the online media display world is one where “demand is large and supply is constrained specifically by qualititative factors.”

As we all know the definition of quality is a very vague one. Sell a brand advertiser and it’s reach and uniques. Test out a direct marketer and it’s all about optimization and direct ROI. The juxtaposition of these metrics creates a very volatile marketplace.

That’s why I like Pubmatic’s announcement this morning. They are essentially serving their clientele (niche publisers) and showing the value of inventory directly through pricing by stating, matter-of-factly through their data review, that the ecpm on small publishers is higher than large publishers.

Pubmatic is further excluding brand advertisers from the review so as to not introduce volatility and noise into their data.

The result? A benchmarking service where a publisher can evaluate their inventory and see if they are performing better, on par or worse than others in their space in terms of performance-based spend.

This helps with a few things:

– understanding if there is more room in their remnant or dr-based inventory

– determining the confidence factor of expecting a similar eCPM the following month

– understanding trends to improve content on their site or migrating directionally towards a certain CPM

– evaluating whether they should go out and hire an expensive sales staff (for brand focus)

At the very least, it shows to publishers that: a) Pubmatic is their advocate and b) there is value for being part of the Pubmatic “network.”

Sounds like a forthcoming convergence here between players like Rubicon Project (mostly remnant-remnant to date), AdRoll, Pubmatic, OpenX and an assortment of others.

Will be interesting….

More on TechCrunch: