The Quest for Perfect Information: How Network Intelligence is Transforming Search — Search on the Edge Track
I have absolutely no witty commentary to take us into this one so instead I’ll introduce our stars.
Our moderator is Ken Miller, Anchor Intelligence. The speakers are Stephen Scarr, Info.com; Bill Leake, Apogee Search; Ted West, LookSmart; and Craig Greenfield, Performics. Looking at that line up, I think I might be in over my head on this one. Why do I always pick the sessions that melt my brains?
Craig Greenfield starts off this session, which is rather under-attended.
In 1998, GoTo/Overture was really the start of this quest for perfect info. Then Google’s launch of AdWords and Yahoo!’s purchase of Overture continued to move it forward. The ability to value display and impression the way we do paid search is the next step toward figuring out ROI.
The search advertiser has four decisions to make:
- Which keyword to target
- Which landing page to use
- Who to target
- And what to pay
He runs through a case study where they were looking toward bringing up the long tail. This is very definitely over my head.
Here, have a slide showing various tasks on a grid of energy versus impact.
I’m so lost right now. I don’t even know what a Bayesian statistic is. I believe his point is that as you become more expert, you will need to expend more energy to see gains but the impact will be correspondingly greater.
You can’t just gather the data — you have to do something with it.
- Crawl, walk, run
- Test, test, test
- Scale matters
The quest for perfect information is really an iterative journey.
Up next to blow my tiny mind is Steven Scarr. He’s going to do so with an excellent English accent, so that’s a bonus.
They focus their SEM on the long tail, which he believes is going mainstream. He thinks it’s odd that Google still limits ad titles to 25 characters despite the fact that search terms are getting longer. As a result, he sees that people are also wasting the opportunity to build longer titles for Yahoo and Bing which offer 40 characters.
People usually click on longer Titles. So, for example, in the slide above the longer Yahoo! results would get more clicks than the Google ones.
Bidding on broad match will negatively affect your campaign because longer terms will often be over the 25 characters limit.
They’ve built tools to try to reduce those errors, fixing the longer terms, shortening them with abbreviations and removing extraneous words so that the titles will fit. It enabled them to “rescue” 32 percent of long-tail keywords that they could actually then bid on, which had previously triggered an error.
Their proprietary tool is Zenya and it’s due to launch summer 2010. It’ll be a subscription model.
Ted West is next.
What’s a search advertising network?
- Keyword-based PPC auction
- Non-proprietary search traffic
- Syndicated network, no owned and operated search traffic
- Diverse search traffic type, media, performance, and value
- Intentional: search boxes, toolbars, domain navigation
- Contextual: in-text Web pages, e-mail messages and social media
The key to efficient market performance on search ad networks is the right keywords matched to the right traffic at the right price.
Selecting the right keywords is built the same way as other sources: Hitwise, Wordtracker, Google Keyword Tool, plus their own internal data.
Delivering the right traffic is where they feel they’re set apart. They have match traffic types, use third-party research and they monitor and optimize traffic sources to meet performance objectives.
They set the right price dynamically. They can segment by type, source, quality and expected performance.
[Is this a sponsored session? These presentations are basically infomercials. There's not a whole lot of meat for someone to optimize their campaign or understand more about search networks.]
Here’s a grab of the most useful slide in Ted’s presentation. It went by fast thus the lack of transcription.
Next up is Bill Leake. He has Tim the Enchanter AND “To Dream the Impossible Dream” on his cover slide. It’s wicked awesome. He encourages the room to tweet more and to mock the panelists if they need to. Bill Leake is my favorite person on this panel.
- We will NEVER get perfect
- Last time he checked:
- rooms are full of humans, not machines
- most of us want more than a Web form
- datasets are dirty, dirty, dirty
- datasets are owned by different corporate functions (and they don’t share)
[Holy cow, he's zipping through his slides. Slow down, Bill, this is good stuff!]
It’s hard to calculate the value of a number one search because it’s a dynamic environment. Blended search changes by geographic regions. Compare Houston SEO company to Chicago SEO Company to San Diego SEO company. There are major differences.
And it’s just as bad going down the funnel as it is going between geographic regions. Natural is a little more research oriented and paid is a little more action oriented, but only a little.
What else dirties the data? Alternate conversion types.
Conversions don’t have to occur on your site:
- You can track calls from search
- Integrating chat to drive conversions
- For B2C retail, add coupons for in-store tracking
- Maintain SEO and PPC traffic codes in your marketing
- A “click” is an early conversion even if it’s not a final conversion
Most people end up buying something eventually. They just don’t necessarily buy it from you. People don’t look for “time sheet software” for fun unless they’re seriously wacked in the head. They’re going to buy something from someone.
It’s easy to be misled on the Web by “Shallow successes” — e.g., cheap clicks or cheap leads.
True ROAS usually varies widely, even with a constant CPL target.
[Seriously why couldn't Bill have done this whole thing?]
Case study: Yahoo! was not generating leads as well as Google. However, looking down the funnel, they discovered that Google cost 151 percent MORE per conversion. Don’t just turn it off until you dig deeper.
Don’t discount re-buy behavior. Make sure your email campaigns are integrated.
Coupon users aren’t loyal.
- Perfection is a journey, not a destination. As Mike Moran says, “Do it wrong, quickly.”
- Think down funnel.
- Try to integrate where you can.