This is a really cool Mashup. Put a image, Create VoiceOver and Supers using drag and drop templates and you can have an instant ad. Here is an ad I created in 90 seconds -
Make your own Attack Ad
Sunday, November 2, 2008
Thursday, January 3, 2008
Saturday, October 20, 2007
Persistent Cookie Death: Is it a problem
All measurements of Consumer Engagement are dependent on snippets of information called Cookies. A Cookie is served by a domain to a web browser.
Cookies introduce a sense of "state" to HTTP, which is a "stateless protocol". This means that the server that responded to the browser's HTTP Request remembers the browser, as long as the cookie is alive.
Cookies are of 2 types from a cookie-life perspective: Session and Persistent. A session cookie expires within typically 30 minutes. A persistent cookie on the other hand has a life of much more, typically days, months or years.
Everybody loves cookies: Is this a problem?
With the granularity of consumer response gleaned with cookies, domains typically several cookies, and it is not uncommon to see over 15 cookies served.
The constraints in this are:
Every Domain can sever a maximum of 20 cookies to a specific browser at any point in time
The browser can have only 300 cookies, (each with a max size of 4 KB - too much)
If this limit is breached, older cookies get deleted. Session cookies are killed first, and then persistent cookies.
If a persistent cookie that I have served to a browser is deleted, I will miscalculate causality. Lets say someone searched for PC and went to www.dell.com. Post that search, the consumer bookmarked the site or typed in directly. If the persistent cookie was still alive I woudl have been able to recognize this visitor as originally from a Search Engine rather than a direct visit, with the objective of crediting points to the Search Engine.
Also, my web analytics tool will treat a repeat consumer as a new consumer, and any personalisation engine that the site www.dell.com may have will be lost, resulting in a sub-optimal user experience.
Is this a problem?
I dont think so. I think 300 is pretty comfortable. I have no emprical data to back this right now, but m going to go count the number of cookies in my machine, and write a blog post about that data soon -:)
But i think there is hope in the 4 KB number. I am unable to check this now, but would think that cookies would be no more 100 Bytes. That means we could relax the constraint on size and make it a total number of cookies subject to not exceeding a total size of 300 x 4 kb = 1.2 Megs.
That is a lot of cookie space. Not something I would lose sleep over.
Cookies introduce a sense of "state" to HTTP, which is a "stateless protocol". This means that the server that responded to the browser's HTTP Request remembers the browser, as long as the cookie is alive.
Cookies are of 2 types from a cookie-life perspective: Session and Persistent. A session cookie expires within typically 30 minutes. A persistent cookie on the other hand has a life of much more, typically days, months or years.
Everybody loves cookies: Is this a problem?
With the granularity of consumer response gleaned with cookies, domains typically several cookies, and it is not uncommon to see over 15 cookies served.
The constraints in this are:
Every Domain can sever a maximum of 20 cookies to a specific browser at any point in time
The browser can have only 300 cookies, (each with a max size of 4 KB - too much)
If this limit is breached, older cookies get deleted. Session cookies are killed first, and then persistent cookies.
If a persistent cookie that I have served to a browser is deleted, I will miscalculate causality. Lets say someone searched for PC and went to www.dell.com. Post that search, the consumer bookmarked the site or typed in directly. If the persistent cookie was still alive I woudl have been able to recognize this visitor as originally from a Search Engine rather than a direct visit, with the objective of crediting points to the Search Engine.
Also, my web analytics tool will treat a repeat consumer as a new consumer, and any personalisation engine that the site www.dell.com may have will be lost, resulting in a sub-optimal user experience.
Is this a problem?
I dont think so. I think 300 is pretty comfortable. I have no emprical data to back this right now, but m going to go count the number of cookies in my machine, and write a blog post about that data soon -:)
But i think there is hope in the 4 KB number. I am unable to check this now, but would think that cookies would be no more 100 Bytes. That means we could relax the constraint on size and make it a total number of cookies subject to not exceeding a total size of 300 x 4 kb = 1.2 Megs.
That is a lot of cookie space. Not something I would lose sleep over.
Friday, October 19, 2007
What is the Value of Internet Audience Rating services
I've been buying consumer response on auctiion mediated media networks for several years now, and I must admit that I know very little about Audience Meaurement services on the net like www.hitwise.com, www.nielsen-netratings.com and www.Comscore.com.
Audience Ratings were important in the old days when advertisers bought OTS, or the Opportunity to Be Seen by an audience.
They're great-to-know and I use them in my presentations. I love the data I see coming out, especially from Comscore, is extremely useful at a very strategic level.
However...
The new world is different. Here "ratings", or measurements of the size of a website's audienceaudience measurement go flying out of the window.
As a buyer of audience response, I am not particularly bothered as to what kind of an audience I expose my message to (within certain limits, ofcourse). I am only driven by the Volumes of Gross response (as in Clicks), Cost per Gross Response, and Conversion of Gross Response into Consumer Engagement (the sum of all the actions taken by the visitor on the website).
In my previous posting, I had talked about Consumer Engagement as the single and only important metric for all brands - with the value in defining this metric as one or more actions on a IP-Enabled device.
With 3rd party ad-serving, Auction Platforms, and real-time consumer-engagement driven advertising, it is easy to operate in an environment where INCOMING DATA, not PRE-CAMPAIGN DATA, are all that matters.
The only algorithm (also called by me as the Google Algo)
In a consumer-engagement-centric model, the advertiser is Audience agnostic, but Action Gnostic. If I can get males in my demographic to view a view and request for a Mobile-Coupon, then metric like CPM, or Unduplicated Reach are mostly academic.
If the need is for scale, then ads are contextually or behaviorally (See Behavioral Retargeting on the 247realmedia.com media Network here) sprayed in more or less random fashion with an intelligent iteration point. Bayesian based logical systems take the input data, the set of beliefs (what do we expect the click to translate into CE across a portfolio of media sources), and keep changing beliefs in a Bayesian way so as to get more and more CE for less and less cost per unit CE.
The beginning of the end for pure CPM Networks:
It is fashionable for media sales reps to say "we dont sell Page Views, we sell audiences". Well, I say, pump your audience into a Auction Platform mediated by a 3rd party ad-server, and give me the results in near real-time, I will spray ads all over your network with some starting solution, and optimize as we go along.
The better I optimize, the more I spend, so help me do this. And the better your System, the more money YOU make! Auction Platforms with intelligent 3rd party ad-serving are simply the most efficient way on the planet to maximize publisher inventory and minimize Advertiser CPCE - Cost per Unit of Consumer Engagement.
As a corollary, Networks which are not CPC and Auction-based, (like 24/7's Global Web Alliance), need to upgrade their inventory monetization infrastructure, or ready to see erosions in Yield over time relative to the market.
More on that later.
Audience Ratings were important in the old days when advertisers bought OTS, or the Opportunity to Be Seen by an audience.
They're great-to-know and I use them in my presentations. I love the data I see coming out, especially from Comscore, is extremely useful at a very strategic level.
However...
The new world is different. Here "ratings", or measurements of the size of a website's audienceaudience measurement go flying out of the window.
As a buyer of audience response, I am not particularly bothered as to what kind of an audience I expose my message to (within certain limits, ofcourse). I am only driven by the Volumes of Gross response (as in Clicks), Cost per Gross Response, and Conversion of Gross Response into Consumer Engagement (the sum of all the actions taken by the visitor on the website).
In my previous posting, I had talked about Consumer Engagement as the single and only important metric for all brands - with the value in defining this metric as one or more actions on a IP-Enabled device.
With 3rd party ad-serving, Auction Platforms, and real-time consumer-engagement driven advertising, it is easy to operate in an environment where INCOMING DATA, not PRE-CAMPAIGN DATA, are all that matters.
The only algorithm (also called by me as the Google Algo)
- DO WHAT IS WORKING.
- SCALE DOWN WHAT IS NOT WORKING.
- TEST NEW THINGS THAT MAY OR MAY NOT WORK
In a consumer-engagement-centric model, the advertiser is Audience agnostic, but Action Gnostic. If I can get males in my demographic to view a view and request for a Mobile-Coupon, then metric like CPM, or Unduplicated Reach are mostly academic.
If the need is for scale, then ads are contextually or behaviorally (See Behavioral Retargeting on the 247realmedia.com media Network here) sprayed in more or less random fashion with an intelligent iteration point. Bayesian based logical systems take the input data, the set of beliefs (what do we expect the click to translate into CE across a portfolio of media sources), and keep changing beliefs in a Bayesian way so as to get more and more CE for less and less cost per unit CE.
The beginning of the end for pure CPM Networks:
It is fashionable for media sales reps to say "we dont sell Page Views, we sell audiences". Well, I say, pump your audience into a Auction Platform mediated by a 3rd party ad-server, and give me the results in near real-time, I will spray ads all over your network with some starting solution, and optimize as we go along.
The better I optimize, the more I spend, so help me do this. And the better your System, the more money YOU make! Auction Platforms with intelligent 3rd party ad-serving are simply the most efficient way on the planet to maximize publisher inventory and minimize Advertiser CPCE - Cost per Unit of Consumer Engagement.
As a corollary, Networks which are not CPC and Auction-based, (like 24/7's Global Web Alliance), need to upgrade their inventory monetization infrastructure, or ready to see erosions in Yield over time relative to the market.
More on that later.
MI5, MI6 and MI 4 - CCE or Cummulative Consumer Engagement
After many years of groping, the Advertising Industry has adopted "Engagement" as the highest order metric for measuring media effectiveness.
In 2005, a body in the US was set up to develop a framework for the media and publishing community to work together with this new metric. It was nicknamed MI4 since four key stakeholder groups were involved: advertisers, agencies, media, and researchers.
Consumer Engagement: Jargon or Math?
While engagement is an intriguing metric, it raises more questions than it answers. What is Engagement? How does one measure it? What is the frequency of measurement? What is the scale, is it ratio-scaled (an analyst's joy) or Nominal Scaled (as a digression, several Old Media Metrics are not Ratio Scaled, lending themselves to analyses that have limitations - the more primitive the data, the fewer and less sophisticated tools the data can be subjected to) ?
Broadly, engagement could mean any action reflecting an experience or touch point with a brand, product, group, or message.
That could involve interacting with an ad, requesting an ad, or even talking about an ad or product.
Note the emphasis on the word "Action". This means that old media metrics, or "Opportuinities to be Seen" metrics do not exist any more.
Consumers requesting ad-hoc updates about the Brand or Product via RSS (Really Simple Syndication), is probably pretty high up in the Hierarchy of Actions. A SMS response to an ad is also an action.
The action need not neccsarily be as an input driven by the marketer. It could also be Consumer Generated, such as one consumer evangelising the product to her friend, or calling in to find out if spares or printing cartridges are available. These consumers are "High-Engagement" consumers, and need a special way to be addresses, relative to consumers who solely react to Marketer driven communcation. Failure to acknowledge and manage these "High-Engagement" consumers can be dangerous for a brand.
Whats happening:
We have lots of research pointing to the fact that an Offline or Old media input is reacted to by the consumer with an action that is exploratory in nature - it could be a call, a visit to the Brand URL, could be a MO (Mobile Orginating) SMS, or it could be a search on Google.
If the response is an online actoion, then it is possible to track Actions by Unique Consumer. WIth Persistent cookies, this can be tracked over the lifetime of that Unique Consumer. (Of-course there are moans and groans about Cookie-Overload, Cookie Deleting, 3rd party cookie blocking etc, but with 300 Cookies, and overwriting of only session cookies on a FIFO basis, persistent cookie death may not be dramatic).
What Advertisers need to do - Today!
With these 2 developments, it is possible to atribute causality to Action, typically my media source. Through using a system of URL 301/Meta Redirects the Browser based tracking system can assign cauality of any action performed online, (including Phone calls, if the mechanism is a Request for a VOIP call-back).
Therotically there are no constraints of either number of actions that can be uniquely measured, or the number of media sources, or the lifetime of the consumer.
Given this model, marketers and advertisers will now rapidly migrate their spend where ideally every single action performed on an IP-enabled media consumption device can be uniquely identified and Post vist actions quantified over the lifetime, leading to a metric that I'll call Cummulative Consumer Engagement.
Cummulative Consumer Engagement with AJAX-based visualisation of All touch points:
Cummulative Consumer Engagement can be quantified in near-real time, and can be correlated on an AJAX interface with Marketing Investments plotted over time. Imagine something like Google Trends or Google Finance, with the Y axis showing the CCE, while the X axis shows time. Roll over on the line and you see the correlation. Marketers, starved of anything but the most primitive metrics will LOVE this.
This does not mean that marketers will stop their TV spends (which account for perhaps 90% of all their spend with advertisers like P&G) overnight. But whats going to happen is more accountability, and the emergence of Consumer Engagement Consulting, a practice that I predict will become a premium huge end consulting practice in the media industry. The mobile phone showed some promise but with clunky WAP interfaces, and mostly SMS interaction, and now SPAM, that promise has died. It will emerge afetr 10 years, and that is why Google is investing in the Mobile Space, but that's the story of another posting.
More on this later.....Rajeev
In 2005, a body in the US was set up to develop a framework for the media and publishing community to work together with this new metric. It was nicknamed MI4 since four key stakeholder groups were involved: advertisers, agencies, media, and researchers.
Consumer Engagement: Jargon or Math?
While engagement is an intriguing metric, it raises more questions than it answers. What is Engagement? How does one measure it? What is the frequency of measurement? What is the scale, is it ratio-scaled (an analyst's joy) or Nominal Scaled (as a digression, several Old Media Metrics are not Ratio Scaled, lending themselves to analyses that have limitations - the more primitive the data, the fewer and less sophisticated tools the data can be subjected to) ?
Broadly, engagement could mean any action reflecting an experience or touch point with a brand, product, group, or message.
That could involve interacting with an ad, requesting an ad, or even talking about an ad or product.
Note the emphasis on the word "Action". This means that old media metrics, or "Opportuinities to be Seen" metrics do not exist any more.
Consumers requesting ad-hoc updates about the Brand or Product via RSS (Really Simple Syndication), is probably pretty high up in the Hierarchy of Actions. A SMS response to an ad is also an action.
The action need not neccsarily be as an input driven by the marketer. It could also be Consumer Generated, such as one consumer evangelising the product to her friend, or calling in to find out if spares or printing cartridges are available. These consumers are "High-Engagement" consumers, and need a special way to be addresses, relative to consumers who solely react to Marketer driven communcation. Failure to acknowledge and manage these "High-Engagement" consumers can be dangerous for a brand.
Whats happening:
We have lots of research pointing to the fact that an Offline or Old media input is reacted to by the consumer with an action that is exploratory in nature - it could be a call, a visit to the Brand URL, could be a MO (Mobile Orginating) SMS, or it could be a search on Google.
If the response is an online actoion, then it is possible to track Actions by Unique Consumer. WIth Persistent cookies, this can be tracked over the lifetime of that Unique Consumer. (Of-course there are moans and groans about Cookie-Overload, Cookie Deleting, 3rd party cookie blocking etc, but with 300 Cookies, and overwriting of only session cookies on a FIFO basis, persistent cookie death may not be dramatic).
What Advertisers need to do - Today!
With these 2 developments, it is possible to atribute causality to Action, typically my media source. Through using a system of URL 301/Meta Redirects the Browser based tracking system can assign cauality of any action performed online, (including Phone calls, if the mechanism is a Request for a VOIP call-back).
Therotically there are no constraints of either number of actions that can be uniquely measured, or the number of media sources, or the lifetime of the consumer.
Given this model, marketers and advertisers will now rapidly migrate their spend where ideally every single action performed on an IP-enabled media consumption device can be uniquely identified and Post vist actions quantified over the lifetime, leading to a metric that I'll call Cummulative Consumer Engagement.
Cummulative Consumer Engagement with AJAX-based visualisation of All touch points:
Cummulative Consumer Engagement can be quantified in near-real time, and can be correlated on an AJAX interface with Marketing Investments plotted over time. Imagine something like Google Trends or Google Finance, with the Y axis showing the CCE, while the X axis shows time. Roll over on the line and you see the correlation. Marketers, starved of anything but the most primitive metrics will LOVE this.
This does not mean that marketers will stop their TV spends (which account for perhaps 90% of all their spend with advertisers like P&G) overnight. But whats going to happen is more accountability, and the emergence of Consumer Engagement Consulting, a practice that I predict will become a premium huge end consulting practice in the media industry. The mobile phone showed some promise but with clunky WAP interfaces, and mostly SMS interaction, and now SPAM, that promise has died. It will emerge afetr 10 years, and that is why Google is investing in the Mobile Space, but that's the story of another posting.
More on this later.....Rajeev
Sunday, September 23, 2007
What is Acquisition Marketing ?
Acquisition Marketing is the pursuit of specific activities that result in actions that are are measurable in near-real time, in units that are typically ratio-scaled, and viewable in near-real-time on a web-interface. Typically these actions are performed by "consumers", and so are called consumer responses. The consumers are people who are browsing content or surfing the the web on an IP-Enabled Device (PCs, Mobile Phones, IP-TVs).
What is not acquisition marketing ?
Let me, big Media Buying company, Godzilla and favourite of the Tyrannosaurus Ad-Holding company do a USD 100 million AD-Campaign and build your Brand Affinity
This is not Acquisition Marketing because:
1. No specific consumer response can be measured.
2. If yes, it is typically not a ratio-scaled response.
3. Response cannot be viewed in near-real time on a web-interface.
4. Results are got post facto, typically long after the aperture of actionability has ended.
The above example is representative of the work done by Old-Media broking companies, companies that are Clueless on the way media consumption, media monetization, and consumer-response exchanges are headed.
What is not acquisition marketing ?
Let me, big Media Buying company, Godzilla and favourite of the Tyrannosaurus Ad-Holding company do a USD 100 million AD-Campaign and build your Brand Affinity
This is not Acquisition Marketing because:
1. No specific consumer response can be measured.
2. If yes, it is typically not a ratio-scaled response.
3. Response cannot be viewed in near-real time on a web-interface.
4. Results are got post facto, typically long after the aperture of actionability has ended.
The above example is representative of the work done by Old-Media broking companies, companies that are Clueless on the way media consumption, media monetization, and consumer-response exchanges are headed.
Subscribe to:
Posts (Atom)