 |
May 17, 2008
|
Aaron Wall has an excellent explanation of why a click costs so much more on Google on average than a click for the exact same term on Yahoo. As he explains, it really comes down to the quality of the syndication partners Yahoo has partnered up with to display ads:
How Click Arbitrage & Dirty Ad Syndication Killed Yahoo! Search Marketing Put another way, a Yahoo! click for mortgage is worth the same $15 that it costs on Google, but it goes for less than $5 because Yahoo! forces advertisers to eat junk traffic too. If Yahoo! virtually killed off their syndication partnerships (at least all but the cleanest ones) their short term revenue might decrease, but their click values & click prices would sharply increase.
A common misperception is that a click for a given term should be nearly equal across all search sites. If someone is looking for a mortgage, is shouldn't make much difference whether they happen to start their mortgage searches by typing the term into Google or Yahoo's search box. And this is true for the most part, outside of some demographic differences between search sites.
What's killing Yahoo's cost per click is their syndication partners. An ad placed into Yahoo's search marketing program appears not only on Yahoo (where you'd expect it to) but also on over 1000 additional sites through syndication relationships Yahoo has built over time. Yahoo splits the value of each click with the syndication partners. Unfortunately, traffic from syndication partners - on average - performs worse than traffic on Yahoo itself. How much worse? Enough to make the value of a click on Yahoo 1/3 of what it might be on Google for the same term.
Is this good or bad? Well, neither, from my perspective. The real-time fluid market that is PPC advertising has managed to determine the market value for terms based on historical conversion rates.
As Wall points out, Yahoo could fix this by giving advertisers more control over where their ads are shown, and for how much. Surely, advertisers would be willing to pay more per click when ads are shown just on Yahoo's own search property. And Yahoo takes home a larger percentage of the revenue generated from those clicks, so it would be a big win for them. Advertisers would likely continue serving ads to Yahoo's syndication partners, but at a more reasonable rate than they're forced to pay now.
Beyond Google and Yahoo, there are quite a few more pay per click advertising networks you could try. It's quite common to see advertisers see some success with one of the big players, then look elsewhere to see if they can find even more PPC traffic - and possibly some at lower costs per click than what they're paying with the top-tier PPC sites. Unfortunately, problems with quality become more and more rampant as you move down through the tiers, so don't get your hopes up too high. Just be sure to have a decent analytics plan in place before throwing money into this type of advertising.
|
May 15, 2008
|
I've been trying to think of a way to illustrate the point I was trying to make in yesterday's post about the value of "micro-short-tails"* and came up with this:
Theoretically, the bars on the map represent three products sold on a website. They vary in popularity with Product A being more popular in terms of both traffic (and revenue, in this case) than Product B or C. The horizontal access represents search terms, so the left side of the graph shows that some terms drive a lot more traffic than terms further to the right. A typical hockey stick style graph (this is based on real data).
The point here is that, while the graph for Product C happens to be smaller than Product A, it's no less valuable when you're looking at just that product.
When someone looks at the value of keywords at a very high level, the terms driving traffic to Product A will appear to be vital, but Product C's terms may go unnoticed or become under appreciated. It's kind of a battle vs war type issue.
The nice thing is that it's often possible to figure out patterns of success. The types of terms that drive traffic to Product A will probably also work for Product B through infinity if applied in a similar pattern of ads, landing pages, offers, etc.
So, while it's true that you'll have less data to work with on really really long tail products due to low search volumes, it's still possible to make some fair assumptions that more often than not will prove correct.
* I think better term for this would be "Micro-Short-Heads" rather than "Micro-Short-Tails"
|
May 14, 2008
|
Brian Carter of Fuel Interactive raises an interesting point about the effectiveness of keywords in the long tail of pay per click advertising. It has long (in Internet time) been thought that advertising on a huge portfolio of search terms will generate better returns because you'll get a trickle of traffic from thousands and thousands of terms at low cost per click AND the traffic will convert at a higher rate for a double gain in returns.
However, Carter explains that this isn't necessarily the case:
The Problem with Long Tail Keywords
But the dirty little secret of PPC is that 95% of your conversions come from 5% of your keywords.
Really.
The others keywords either
* Don't perform (100 clicks and no conversions), or
* The clicks roll in so slowly that you won't have the statistical confidence to delete them until the year 2112 (yay, Rush!).
As I said, interesting points.
While 95% of conversions may come from a small sub-set of a keyword portfolio, that doesn't necessarily mean that the rest of the portfolio is underperforming. The rest of the terms may not deliver as many conversions, but that's not really a measure of performance. Return on investment would be a better measure.
But more importantly, I think the biggest concept that could be misunderstood here is what I'll call "micro-short-tails." By that, I mean terms that get relatively few searches, but are still clearly short-tail terms when looked at on a page by page basis. For example a retail site could have thousands of products in inventory - some of which are relatively obscure. On a page by page basis, it's pretty clear that product-names and product-IDs would be considered short-tail terms while on a site-wide basis they would look more like long-tail terms.
In many cases, patterns of "micro-short-tail" terms can be generated and properly targeted to truly relevant pages.
Is this long tail or short tail? In my opinion, it's all relative.
I think Brian and I would agree that measuring what's working and building upon successes is the key to running great pay per click campaigns regardless of how different strategies are defined.
 |
May 13, 2008
|
Today I had Comcast's new DOCSIS 3.0 Business Class service installed at a home with 5 static IPs. The install went fairly smoothly (we had a bit of a scare in thinking we couldn't get over 2Mbps but it all worked out in the end.) The nice thing about Business Class is that you get a Cisco 1800 series router, but the strange thing is that they still use the Scientific Atlanta router too (why not just use the 1800?)
One of the problems with Comcast's new DOCSIS 3.0 service is that it is simply too fast. At 50Mbps download speed (and 5Mbps upload speed) your service will be going faster than most web sites can deal with. This means it is very hard to do a proper test to see if you're actually getting the 50/5 service that you paid for. Think the SpeakEasy.net speedtest can help? Depends on how close you are to a node.
The solution is simple. If Comcast had some sort of super-secret internal speed test page that would allow you to see the speed that the Comcast network is sending you before it hits the public Internet, that sure would be helpful, wouldn't it?
Take your new DOCSIS 3.0 connected computer, open a web browser and point it to http://68.87.178.6 to get the Comcast internal speed test page. This will test the speed of your cable router inside of the Comcast network allowing you to see the raw speed as it goes to your house. While this may not be a local node near you, it is on the Comcast network and should give you better results than any external test would give you. I know that this address is what they are using in Minneapolis, MN and your area may have a different page. Now don't sit on your computer and keep refreshing that page. Every time you use the test you are trying to max out your router, and as such you are putting additional strain on the Comcast network (which in the end could make it SLOWER). Don't come crying to me if you keep hitting the page over and over again then get a nasty call from Comcast. Use it to see if you're getting your advertised speed from Comcast, or at least something close to that (it may be a bit under still, but it should be close).
Of course it is worth mentioning that fundamentally all speed tests are flawed and will never be able to give you a 100% accurate rating of your actual Internet speed, but this is at least a better option than trying to guess at which node is closest to you or has the least amount of traffic/latency/issues.
I've been disgusted with just how bad airports, in 2008, are at accommodating the needs of business travelers.
As I understand the world of travel, business travelers are the ones who pay a lot for their tickets. I also get the impression that most major airport budgets come from landing fees, which are largely derived from planes carrying business travelers.
With that in mind, doesn't it seem strange that airports force business travelers to sit on the floor because they don't put outlets near chairs?
Weary Business Traveler Charging His Blackberry
Is this how you treat guests in your home? You can plug your computer in over there and sit on the floor?
Some airports are providing Internet kiosks like this one for travelers:
In this case, it includes the option to pay for power.
A better option is to unplug the Internet kiosk that nobody uses and plug in your computer to the wall.
One other option I've found for the truly power-hungry. Tell the gate attendants that you have a special condition. After entering the ramp, set up camp at the outlet within boarding ramp.
Give it a try and let us know how what kind of reactions you get.
|
May 12, 2008
|
Should you host your own blog? If so (or not) what are your options and why should you choose one over another?
These are common questions from people who are considering getting into blogging for business or something beyond blogging on their MySpace or LiveJournal account.
Based on my experience, here is how I'd help guide someone to the right solution for their needs today:
This will surely change over time as new players enter the market, current players slip behind, etc. so please keep in mind that the recs here represent here and now and will probably look ridiculous in a year or two .
But, with that in mind, the important thing to remember is that any new blog platform will have to make it easy for currently bloggers to transfer their blogs if they hope to gain an audience. If that's the case, most of the blog platforms mentioned in this diagram should be transferable. Stay away from MySpace if you think you'll ever want to republish your posts somewhere else.
|
May 10, 2008
|
Jamie Thinglestad (Dow Jones), Robert Stephens (Geek Squad), Dan Grigsby (Unpossible), Douglas Olson (Microsoft), Matthew Dornquast (code42), Michael Gorman (Split Rock Partners)
How to attract software companies to come here?
Grigsby: MN has a long track record of creating products for large companies, including Dow Jones, Microsoft, and Oracle. We should be embracing this. Should software be the business or software for business?
Let's take developers out of the banks and put them in start-ups.
Thinglestad: Question he hears on the coasts: Can you really build a development shop in Minneapolis?
You can build a core team anywhere. Talent in the Twin Cities is "pretty high." "The Hollywood Factor" is where struggling developers go to where the jobs are. Great developers can work anywhere.
It's not cheaper for talent in MN except maybe at the entry level.
Olson: If you're looking to save labor costs, you'll go to China, India, or Vietnam rather than Minnesota. Getting paid the same in MN vs California will give you more to work with here.
Didn't think he could hire the people he needed for AuthorWare here in MN, so he moved the company to California. He ended up hiring students from Big-10 schools who had move to CA after moving the company there. After moving the company back, he found employees who had relocated to MN from the coasts.
Stephens: We're the Russia of America. We don't have the distractions of the beach and mountains.
Gorman: We must always remember that we're in a global market so the most compelling ideas and teams will win internationally.
Gribsby: At a certain point in your life, you can make a decision to be a software entrepreneurialist. We need a denser concentration of software development business people.
Olson: U of MN is not a top school for software development, which makes it difficult to find top local talent.
Stephens: Geek Squad is the janitors of the IT industry. Within large companies, people don't try out the $500 ideas because it costs $100,000 to get permission from the IT department. Big companies need to understand how to prototype faster. Have sandboxes. We need a "corporate API" where it becomes easier for developers to show what's possible.
Dornquast: Uses the term "Silicon Prairie" to refer to software development in Minnesota. Thinks, psychologically, that Minnesotans are risk-averse compared to people on the coasts. Get started: try out your idea. If you have something you're passionate about, do that first. Don't waste time working on risk-averse applications.
Olson: On entrepreneurial software development: "It doesn't take a lot of money to write software. It takes guts, boldness, and confidence." On the U of MN: We have to be able to raise our children around those who are the best of the best.
Stephens: TPT is starting Make:TV this fall. A Maker Faire inspired show sponsored by Geek Squad.
Dornquast: Toughest thing he had to do in his start-up: create a product brochure. Luckily, his wife works in marketing. Find the right people you need to help you succeed.
Gorman: On smaller investment opportunities due to lower costs of Web 2.0 startups: Angels and VCs will both fund Web 2.0 companies on smaller investments. However, not all investments will be built on $500,000 and sold to Google or Yahoo for $20 million. Angels provide a critical component to the development community.
|
