JohnDeMayo.com: Internet Advertising Insight

Value of a mortgage lead by state (chart)

Some interesting data on the relative value of a (refi) mortgage by state.  Aggregated from a few sources.  I will leave you guessing as to the designation of most of the states, but this should give you an interesting perspective on the huge variance in value looking at just one dimension.  See an older post on what other dimensions are important here.

Also see:  http://www.johndemayo.com/johndemayocom/2006/08/of_times_a_mort.html

 

 

Traditional or bidded sales model II

    As a follow-up to some questions sent via email in regards to the attributes of a model that dictate possible success via bidded environment, there are definitely important dimensions other then the # of bidders able to bid on a potential lead.

    When I was at Advertising.com and we were starting to tinker with allowing advertisers to adjust their price online without talking to a rep at will (and therefore receive a corresponding change in volume on the network), I was pessimistic on the bidded model because at first at least, we lacked transparency in regards to what a price change was really doing.  While advertisers understood the basics of higher price = more exposure, without a transparent marketplace of exactly how much volume is anticipated, and ideally some explanation of what was occurring specifically to change their volume, the effects off the bidded model were somewhat muted.  Transparency in regards to the expected change in volume, and ideally exactly what is causing the change in volume, is crucial to reaping more of the benefits of a bidded environment.

    Similarly the “winner distribution” is essential to creating a bidded environment.  In a model where the highest payer takes all of the volume, you will normally have too few bidders competing to raise prices optimally, and clients will get frustrated with the extremely hard to control and sporadic volume, as they capture all or none of the volume.  One of the simpler models for bided environments where you might expect 5 or 10 companies to compete for a given lead (which will only be sold to one party) is to simply give the highest bidder ½ of the volume, the second highest bidder ¼, the third highest bidder 1/8th, etc.  With this simple methodology you give 99% of your leads to your 6 or 7 highest bidders, keep multiple bidders active, and allow bidders at the bottom of that tail to test a few leads, without leaving too much money on the table.  While not as smooth of a distribution as cited above, you can see probably guess what the distribution of volume for bidders on a particular search term might look…..where the decay after the first few spots slows, and where there are big jumps in volume depending upon whether you are on the first page of results or not.  I suppose there’s probably some math out there that could help with predicting the optimal “winner distribution” for a market with a given set of attributes.

    So if you have more then a few bidders desiring a particular good, you can provide transparency into what a change in price will do in regards to volume and why, and you think you have a reasonably optimized “winner distribution” model – put some of your leads up for bid, and let us know how it goes.  I’m seeing more and more companies try it, hopefully we’ll see more implementations that make sense become successful in the future.  The model certainly did wonders for paid search.

._-addition-_.

    Greg Yardley points out that it's not just a transition to a bidded model, it's a transition towards automation and transparency.  It's also not just about leads, it's about buying/selling all of online media.  And he's right.

Senior engineers needed for sizzling hot start-up

Senior Search Engineers for hot Silicon Valley startup (Palo Alto)

We are a profitable, top tier venture-funded, pre-IPO

Silicon Valley

company. Our Internet business is growing at the rate of 60% each month. Our investors were early advisors and research collaborators at Google, and the management team are veterans of several online advertising, marketing, and search giants such as Shopping.com, Advertising.com, Amazon, and Yahoo. We have PhDs and Faculty from leading universities such as Stanford and Cornell working for the company.

 

We are looking for a few superstar search engineers to join our core team and play a critical role in enhancing and building our technology foundation. The ideal candidate is a smart, productive self-starter who has had startup experience and understands the demands of software development in a startup.

 

You must have prior work experience with:

- Search engine development (crawling, indexing, searching)

- Information retrieval

- Clustering and classification technologies

- Natural language processing, machine learning, data mining.

 

If you meet these requirements and would like to create robust, high performance search software as part of a great team, please send your resume in plain text format (no attachments please) to clnetinc@yahoo.com. We would love to hear from you.

 

Senior Java Engineers for hot Silicon Valley startup (Palo Alto)

 

We are a profitable, top tier venture-funded, pre-IPO

Silicon Valley

company. Our Internet business is growing at the rate of 60% each month. Our investors were early advisors and research collaborators at Google, and the management team are veterans of several online advertising, marketing, and search giants such as Shopping.com, Advertising.com, Amazon, and Yahoo. We have PhDs and Faculty from leading universities such as Stanford and Cornell working for the company.

 

We are looking for a small and select group of superstar engineers to join our core team and play a critical role in enhancing and building out our technology foundation. The ideal candidate is a smart, productive self-starter who has had startup experience and understands the demands of software development in a startup.

 

You must have prior work experience with:

- Web application development

- Java, J2EE, XML, Apache, Tomcat

- MySQL

- Linux

 

Experience in one or more of the following areas is a plus:

- High performance distributed system development

- Search technologies

- High performance math software

- Innovative data structure and algorithm development

- Scripting (Perl, Python, shell)

- C++

 

If you meet these requirements and would like to create robust, high performance Java Web apps as part of a great team, please send your resume in plain text format (no attachments please) to clnetinc@yahoo.com. We would love to hear from you.

Typepad issues

Typepad, the service that powers this blog, suffered some downtime from Thursday evening through Friday evening or so, and they temporarily lost some data.  As a result, my last post was temporarily lost.  It has now been reposted so if you were trying to view that individual page and got a 404 error, the post is back as you can see.  As a result however the "published on" date was changed until today.

I like the ease of the fully-hosted typepad service, but they have definitely been having some hiccups lately.

Sales outlet: Traditional rep or bidded environment model?

As google gains some traction selling site-specific graphic ads via a bidded model, there is increasing awareness of the power of the automated and bidded model as a method to generate a lot of ad sales revenue, without a lot of headcount in the sales department.  The hope is that a bidded model increases the price advertisers pay as they bid each other up towards their respective break-even points, and your sales cost are lower then a traditional rep sales model.

Lead generation and related companies appear to be taking increasing notice of this model as something they might be able to apply to their sales channels, and I have seen a number of stabs at moving towards this from "lead gen" companies.  I've even heard a struggling company describe this as a bit of a magic bullet, thinking it will magically increase average lead sales prices and enable them to lower sales costs, in a market where the bidded model simply isn't feasible.

More thought needs to be given to the sales model in particular lead generation verticals, and I expect to see more and more blended models, where the line between traditional person-to-person negotiated sales with firm price ranges is blended with a google-like bidded model.  The short answer is the more companies that demand a particular lead, the better the bidded model should work.  When there are only one or two "demanders" for a given unit of supply however, the bidded model is far from optimal and a more traditional approach with reasonable hard pricing floor will net a greater revenue per lead.

Consumating

Congrats to friend Ben Brown on his sale of Consumating.com to CNET.  While a bizarre acquisition for CNET, I hope all involved have fun in the continued development of this one.

http://www.corante.com/dating/archives/2005/12/07/cnet_buys_consumatingcom.php

http://www.techcrunch.com/2005/12/07/cnet-acquires-consumating/

Data sources needed

Hoping someone can help me find an appropriate data source, to help with recency updates to our keywork grouping /bidding / clustering algos.  Budget of up to a few grand per month for the right provider(s).  Need one or more of the following:

1)  A search engine, who can give us real-time or next day access to a raw search query log.  Not monthly, not rolled up in any fashion.

2)  A software company, adware company, or ISP, who can provide the above, OR even better simply a list of every URL requested by their users (WHich would naturally include search engine queries inclusive) again next day or real-time, not rolled up.

Happy to pay a referral fee for someone who can point me in the right direction.

Visitors by month

This blog's content, I hope, relates almost entirely to internet direct marketing.  And readership to this site is up, at least according to this chart.  My only down month, September, I didn't publish a single post.  With that said, I left the scale off of this chart on purpose -- my readership isn't huge -- nor even large -- compared to so many others.  Because I focus on a fairly narrow segment of the industry, and because I try to stay away from web2.0 and M&A rehashings, my audience is narrow, and my posts are less frequent then those of many other blogs.

With that said, I'm most excited about the core group of half-a-dozen or so individuals who are interested in working with me on a new project, one that I hope will enable us to be the leading direct response internet advertising blog.  Not that there is a lot of competition in that narrowly defined of a space, but still, it's an area that I know, love, and have lived and breathed for a long time -- and I long for more industry-wide communication and knowledge-sharing.

So, if you have any interest in contributing your thoughts on this space, either in a regular or on a  every-once-in-a-while basis, drop me a line.  It makes sense that this type of content should be aggregated on a content/vertical basis, not just on an individual basis.  And if our readers like what you have to say, we are more then happy to point a large link to your personal blog, corporate homepage, or whatever you choose in this regard.