Interview: George Michie on Paid Search
Friday, March 12th, 2010
Not long ago I had the privilege of teaming up with George Michie of the Rimm-Kaufman Group for an interview on ecommerce issues. George has graciously reciprocated to offer his expertise in paid search. George is a well respected thought leader in search marketing, serving as co-founder and CEO of RKG, member of Google’s SEM Advisory Council, professor at Market Motive, columnist for Search Engine Land, frequent speaker at ecommerce and search conferences and has worked for top retailer Crutchfield Corporation. You can follow George’s blog here.
Linda: We’re seeing a lot of changes with Yahoo and Bing merging. Does that make things easier or more complicated for advertisers?
George: We think it will help search advertising, at least in the short term. The richer data set afforded by combining the Bing and Yahoo.com traffic will be a boon to those who know how to exploit the long tail, and having one less set of copy changes to make won’t hurt either.
Long term, is the industry better served having one larger competitor to Google or two smaller ones? That’s a tough question. To the extent that Google acts as simply a marketplace where advertisers compete against each other for placements, one could argue that a Google monopoly poses no real risk. On the other hand, given an absolute monopoly, who’s to say that Google will stick to the auction model? If negotiated placements yielded more revenue per impression and greater profits, they’d drop the auction in a second. I don’t think that’s likely to happen; I think the reason for Google’s success lies in the fact that the ads are cost effective when managed properly, and moving away from that model makes them just another portal.
Linda: It’s rare to talk about paid search strategy and escape the topic of the “long tail.” As you elude to, those who know how to work the long tail have an advantage. [Reader note: “Long tail” queries refer to searches that happen only a handful of times per year (maybe only once) can make up the lion’s share of clicks and sales.] What is the best way to capture the long tail? Is the broad match type with a thorough negative keyword list the answer, or should you try to predict the tail with keyword research and exact match terms?
George: I’ve written about this a good bit recently. The importance of the tail does vary a great deal from advertiser to advertiser so it’s not quite true that it matters tremendously for everyone. However, when it does matter, there’s no question about it: broad match by itself is not the best solution.
The right way to approach the tail is to build out the keyword list thoroughly and carefully, with attention to landing pages and copy to maximize both CTR and conversion rates. Then, with sophisticated bid management and flexible attributes applied smartly, the real power of targeting bids to the differentiated value of the traffic takes over.
We like using broad match as a catch-all with lower bids on the broad matched versions than the exact matched versions. The keyword list can never be perfectly comprehensive, and broad match is tremendously valuable for catching word orderings and typos that you’d never catch with exact or phrase match.
Linda: Keeping up with the changes in search engines and advertising program capabilities is a full time job in itself. It’s almost as if when you take a week’s holiday, you’ve surely missed a major announcement or change to the way things are done. The result is do-it-yourselfers may be relying on outdated books or blogs for guidance in managing their PPC campaigns, and even the professionals can be behind the times. In your experience, what’s the most outdated tactic that is still in use by too many advertisers?
George: Things do change all the time but in many senses the fundamentals haven’t changed, and what surprises me is the number of folks who still don’t seem to know the basics.
The fact that so many folks are still bidding by position just amazes me. In fact most of the platforms available for rent and used by many agencies are nothing more than position crawling systems. Even when turned to “efficiency targeting”, they find efficiency by hunting and pecking for the right position, rather than measuring the value of traffic, bidding what makes sense, and letting the position fall where it may. Position crawling guarantees inefficiency and lost opportunity, but even though careful studies have proved that the value of traffic is largely independent of position, people still use position crawlers.
Linda: Surely ignorance is not bliss! That reinforces the importance of having a search manager that really knows what they are doing. Often paid search is outsourced because the retailer does not have the in-house expertise to manage or even understand a campaign, which leaves the program without much oversight. How can you verify your campaign is being managed well? When do you know it’s time to fire your agency?
George: The devil is in the details. I outlined a process for a top-down evaluation of a paid search program. It’s imperative to periodically take the program apart to make sure it is being managed well. Too many folks assume that because they’re using a well-known, expensive agency that their program is therefore well-managed.
Certainly, you shouldn’t have to micro-manage your agency; if you do, what’s the point of outsourcing? But if periodic, detailed evaluations of the program reveal significant problems, the solution isn’t micro-management, it’s finding a competent agency.
Linda: I suppose when clients comes to you after working with another agency, there’s a certain degree of “reworking” you need to do. In your experience, what is the biggest (or most common) mistake that online retailers make in paid search?
George: The biggest mistake is also stunningly common, it’s failing to separate results of brand search and competitive search. Particularly for companies that do a great deal of offline marketing, brand search represents a fundamentally different subset of customers, those who are proverbially ‘walking through your front door.’ Crediting these sales to paid search efforts leads to tremendous overspending in competitive search.
Advertisers mistakenly believe that since the overall cost to sales ratio is reasonable, they’re spending money wisely. If we think instead about the law of diminishing returns and the question: what happened to the last $1,000 we spent and what would happen to the next $1,000? We’ll end up seeing that those answers are all found in the competitive non-brand search program, because that’s where all the incremental spending happens.
Linda: Speaking of keywords, it’s rare to talk about paid search strategy and escape the topic of the “long tail.” As you mentioned earlier, those who know how to work the long tail have an advantage. [Reader note: “Long tail” queries refer to searches that happen only a handful of times per year (maybe only once) can make up the lion’s share of clicks and sales.] What is the best way to capture the long tail? Is the broad match type with a thorough negative keyword list the answer, or should you try to predict the tail with keyword research and exact match terms?
George: I’ve written about this a good bit recently. The importance of the tail does vary a great deal from advertiser to advertiser so it’s not quite true that it matters tremendously for everyone. However, when it does matter, there’s no question about it: broad match by itself is not the best solution.
The right way to approach the tail is to build out the keyword list thoroughly and carefully, with attention to landing pages and copy to maximize both click through and conversion rates. Then, with sophisticated bid management and flexible attributes applied smartly, the real power of targeting bids to the differentiated value of the traffic takes over.
We like using broad match as a catch-all with lower bids on the broad matched versions than the exact matched versions. The keyword list can never be perfectly comprehensive, and broad match is tremendously valuable for catching word orderings and typos that you’d never catch with exact or phrase match.
Linda: Yes, sometimes the queries that trigger your ads can be very irrelevant (even humorous!) We would get searches like “granny cart” triggering our ecommerce software ads at Elastic Path, for example. When you add them up, there truly is a long tail of negative matches that can really dilute your click through rate.
Linda: If you’ve added sufficient negative matches and a keyword still suffers from low click through or abysmal conversion, Is it ever a good idea to delete the keyword? What are some strategies to turn a “dog” keyword into a “star”?
George: I once went so far as to say “there are no bad keywords, only bad bids.” That might have been a bit overstated, but my point was simply this: there are very few keywords – assuming they’re reasonably targeted — that draw in traffic of zero value. The value might be small, but it’s rarely zero. So, first and foremost is the notion that it isn’t a matter of finding keywords that “work” on the first page, and turning the others that “don’t work” off. Bid the appropriate amount and take whatever traffic you get.
Now, it is possible to improve the value of the traffic on a given keyword by identifying negative associations to knock out the wrong traffic, and by making sure the copy qualifies traffic as much as it can. If you sell “desk chairs”, but yours start at $350, where Staples has “desk chairs” for $35, you may find “$350 + for quality office chairs” helps steer the bargain hunters elsewhere.
To a large degree, paid search is dominated by mass marketers. Keywords that speak more to mass market intent can be tough for niche marketers who don’t have selections that appeal to the mass market.
Linda: Ad copy, like any copy, can benefit from split testing. What have been some of your most effective PPC tests? Are there any “rules” to live by?
George: There are very few rules that apply to every advertiser, so maybe one universal rule is to test assumptions! In truth, what may be a universal rule is this: never assume that the traffic on Google.com will perform the same as traffic on Google’s syndication network – even more true for Yahoo and its syndicates. Separate campaigns and measured bid differentials for the engine and its partners almost always delivers positive results.
We have some very exciting results from Yahoo’s new syndication bidding settings that should yield more revenue for our clients, and for Yahoo.
Linda: Sounds juicy! For my final question (you cannot interview a guru without asking something to this effect): where do you see the future of paid search going?
George: The two biggest changes on the horizon are likely to involve the ad formats and how paid search integrates with other marketing efforts.
Google is testing and testing to find formats that generate more revenue per impression, and they may find something. I don’t think product images are going to be it, though. Someone searching for “Pet Supplies” isn’t likely to respond to any particular product image. That said, giving marketers and agencies flexibility to design appropriate display ads for different levels of search depth might work. Providing a plus-box like expansion to the next level of depth, rather than all the way to the bottom with products, might be the best solution.
The ability to parse credit between multiple ads within a search program and across multiple marketing programs will become imperative. We’re launching an attribution management solution for our clients by the middle of March, and I’m sure other agencies will do the same. The better we understand what advertising dollars truly drive sales, the more efficiently we can allocate resources.
It keeps getting more complicated, but that keeps it interesting for us geeks.
Linda: It sure does! George, thanks again for sharing your brain. And now, we open up the floor to questions…
Original post by Linda Bustos
This post was originally published in August 2008 and is part of our Get Elastic rerun series.
Wanna do some extremely cheap (free) and fast market research? As lovely as 
This is Part 3 of our PPC Myth series. Please check out 
Because 
Welcome to PPC Myth week! Today is the first installment of a 3 part series challenging common misconceptions about search marketing and analytics.
If you use the broad match type in PPC advertising, negative matched keywords are essential to keeping your campaigns under control. But are you using negative matches to their full potential?
Last summer we did a collaborative post with fellow Vancouverites VKI Studios called 
It never ceases to amaze me how much Amazon packs onto its product pages. In addition to the products it sells, Amazon product pages include banner ads, links to the product from other sellers (Marketplace), forum links, Listmania, more banner ads, Sponsored Links, Customers Who Bought Also Bought, Better Together, reviews, customer tags, Amapedia…
Last January we held a webinar called 
Last week Google introduced a new twist on its 
This is the final installment of our PPC copywriting series. Part one discussed the importance of 
Subscribe to our feed