Richard is Founder, President, and Chief Gooroo for AdGooroo, and is a long-time internet marketer over 15 years experience in technology and advertising management. He was previously a technology executive at Publicis Groupe/Leo Burnett.
He has a BS in Computer Engineering from the University of Illinois and an MBA in Entrepreneurship and Technology Management from the Kellogg Graduate School of Management at Northwestern University. Richard is a regular speaker on search marketing topics, is a certified expert in both email marketing and conversion optimization, and is the author of “Mastering Search Advertising – How the Top 3% of Search Advertisers Dominate Google AdWords“.
Eric Enge: Tell us a little bit about the book and what motivated it.
Rich Stokes: The book started out as the result of a study that we did. We started in December, and we surveyed about eighteen different industries. And what we were hoping for was to just compile a list of interesting statistics, and what advertisers were doing in the adverts market place.
What we found really surprised me, that fewer than 3% of advertisers were getting 80% or more of the available traffic from Google. That is pretty much what motivated me to do this book.
Eric Enge: It seems to me that there are just a lot of people who structure and build their campaigns very inefficiently, and that’s probably substantive part of the explanation for why 3% of the advertisers get 80% of the clicks.
Rich Stokes: That’s right.
Eric Enge: Okay great. So let’s talk about some of the things from the book in sequential order. For example, how about some common keyword mistakes?
Rich Stokes: The biggest mistakes that I see these days are people just dumping in thousands of terms. They don’t really pay too much attention to the relevancy of those terms. So, you can get dinged for that pretty big on Google these days. They are not only looking at your click through rate for any individual term, they are looking at your click through rate for the campaign as a whole.
If you have a lot of really off-target keywords, it doesn’t penalize you too much, because they have poor quality scores and don’t end up being shown very often, if at all. It’s those mediocre keywords that kill you. The ones that don’t do horrible, that Google doesn’t have to be as active in to raise in bids, and are not really high enough to standout above the industry average.
Taking your competitors and your own campaigns into account, that’s the really important thing. So, the number one keyword rule is to focus on keywords that are extremely relevant to your campaign. You want to focus all your efforts exploring those keywords to the maximum percent possible.
Eric Enge: Right. So, the big issue that you highlighted essentially is that the mediocre keywords are probably going to get lower click through rates and drag up the cost of your click throughs.
Rich Stokes: Yes.
Eric Enge: Right. So, click through rate gets dragged down, and the cost per click goes up.
Rich Stokes: That’s right
Eric Enge: So, what’s the right way to choose keywords then?
Rich Stokes: Well, you really have to take a disciplined approach to it. So, one of the things that I recommend in the book is to use a wide variety of tools. Now, everybody knows of the free tools out there, and there are good ones. I would say, I think the biggest free tool out there is Google’s own keyword suggestion tool which is built right in the AdWords. It’s a great tool; but it doesn’t return a lot of results though.
It will tend to be between two and three word phrases. They give you a lot of bogus ones, and today it’s a computerized algorithm that’s making the suggestions. So, that’s just something you have to live with.
Wordtracker is really useful to building out your initial keyword list. One of the key differentiators in the market place is that there is good data out there. You have to pay for it, however, because that data is not free to the company that’s providing it to you.
Fewer people are willing to pay for keyword research, and therefore you have fewer competitors that know about those keywords. So, you could win on two fronts there. Another real resource for keywords that requires a little bit of programming knowledge, is to go into the APIs of the different search engines.
That’s something I haven’t covered in the book because it’s beyond the capabilities of the average search marketer. But, if you have programming resources at your disposal, getting direct feeds into Google, Yahoo, and especially Microsoft live search APIs, is a really smart thing to do.
Eric Enge: What data are you getting through those APIs that’s helping with the keyword selection?
Rich Stokes: Well, keyword suggestions for one, but also keyword traffic estimates straight from the engines themselves. In Yahoo and Microsoft’s case I don’t believe they exposed the traffic estimates directly. I had heard that Google has started to expose them to the AdWords interface, but I haven’t confirmed that myself, though.
Eric Enge: And, I think another area that’s an interesting way to come up with keywords is when you have broad match type keywords, you get a lot of clicks obviously on things which aren’t the exact phrase you bid on. You can analyze the search phrases that people are coming in and then take some of those, and put them in as separate keywords.
Rich Stokes: Exactly. That’s a great practice for tuning your campaigns.
Eric Enge: Talk a little bit about the customer life cycle model.
Rich Stokes: What that model says is that regardless of your industry, you have at least three major constituencies of web traffic who are searching for you. One of those three groups is the browsers, those are people who are not really sure what they want, but generally are looking to get educated and for some more information on a particular subject.
You also have shoppers who may know something about the product. They probably have identified a need, but they don’t know what their options are for filling that need. And then you have the buyers who have identified their need. They most likely have the product or the service identified, and they are looking for specific information. In particular, pricing is a very big one with that group.
So, the differences in these three groups really tell you a lot about how you structure your keywords, your keyword groups, as well as your ad campaigning around the pages. Buyers are less likely to purchase, and they are looking to get educated. You don’t want to spend a lot of money on that group. You also want to focus your marketing efforts on educating the customer, in some sense setting the bar for their expectations.
IBM did this. In fact, we put the book out there to educate the market heads about search advertising the way that the big boys are doing it. Obviously, the big boys always get customers. We hope to not only show people how to get better results from search marketing, but in the process they also begin to see the value of our tools. So, our book is a good example of something aimed at browsers; but shoppers and buyers are different stories. So, shoppers are typically looking for comparisons and reviews, whereas buyers are looking for specific models. They are looking for reviews on specific products, pricing, delivery options and all sorts of things. So, the traffic is there and I think there is a spectrum.
With the browsers you are doing the soft sell, but with the buyers you are doing the hard sell. You might want to put things like your pricing exactly in your ad for instance. You might want to say “hey, free delivery if you order today.” You don’t want to spend a lot of money on browsers, but you probably want to spend top dollars for the buyers, because those are people most likely to buy something from you, and shoppers are typically somewhere in between.
Eric Enge: What about setting up ad groups?
Rich Stokes: A naive way of structuring an ad campaign would be to take every single keyword or close grouping of keywords, and then create three ad groups for each one of them. One would be browse bucket, one would be shop bucket, and the third would be for the buy bucket. Then you would tailor your bids and your ad copy in your landing pages for each one. It’s a lot of work, but the book talks about how to shorten that process.
Most of the time keywords would only fall into one bucket. You can generally, though not always, tell which behavioral bucket a keyword falls in based on two things. One is the length of the keyword phrase in terms of the number of words. And, two is the specificity. So, if I ask you like what category ‘antivirus’ falls into, that would obviously be.
Eric Enge: Pretty general.
Rich Stokes: Yes If I ask you what bucket would be a “Magnavox DVD 100 television, a nineteen inch television set” be in?
Eric Enge: They would be buyers, when you specify a model number you are not a shopper, unless you add the word ‘reviews.’
Rich Stokes: Exactly. That word reviews I think is a good case of where you are not sure what the phrase might be. If they add a review they might be looking to buy, or they might just be looking to compare different models, in which case they are shoppers. The good news is you don’t have to have three keyword groups for every group of keywords in your campaign closely related by the keyword phrases. You generally will need one and in some cases you may need two coming down to the trial and error. You can even put more scientific testing around that to figure out which is the best way.
Eric Enge: Let’s talk about power keyword matching.
Rich Stokes: Keyword matching is a great thing; most people use broad matching only. And, as far as it goes, that’s the easiest, lowest effort thing that you can do. But, adding in different match types is almost as little effort, and the payoff is absolutely huge. Let’s say we take the Magnavox television as our keyword phrase.
Our naive approach would be to put this phrase, and related phrases, in their own ad group, and create some ad copy and a landing page for that particular keyword. With two little tweaks you can dramatically increase the click through rate, while simultaneously lowering the cost per click of that group. And, that is simply adding the two additional versions of the keyword phrase of Magnavox television using exact matching, and phrase matching.
It doesn’t always work for every keyword, but generally for two or three word combos where you have a pretty good chance of getting exact match, your click through rate can absolutely skyrocket. In those cases where you get an exact match you stand a very good chance of getting your ad pushed into the premium positions on the search results page, and getting a correspondingly high click through rate.
Setting this up takes a very short amount of time. I personally have an excel spreadsheet that I have rigged up that does it for me. I have the broad match version in the first column, and then my second column adds brackets around it, for exact match. Then, my third column has quotes around it for phrase match. I can take that and I paste that into AdWords and then I get the job done very, very quickly.
Eric Enge: Right. And, it’s basically as you’ve explained, that the gain you get out of that is that exact matched phrase effectively improves your position, because when the exact matched keyword is typed in, it’s a stronger match.
Then, correspondingly with the phrase match it has the phrase in it. It will do better than a broad match that matches up with the same thing.
Rich Stokes: You got it, exactly. The other type of matching that I’ll touch on is negative exact match. If you have broad keywords that generate a lot of traffic and you can’t bear the thought of losing that traffic, then the negative exact match is really a great way to go. I have used this in the past to target keywords like Spyware. What the negative exact match does is show my ad on any longer phrase containing Spyware or anti-Spyware, but doesn’t show it if somebody just types in that one keyword. That ends up filtering out a tremendous amount of traffic.
90% to 95% of people that type in the keyword Spyware into Google do not buy anything. So this effectively filters the traffic you don’t want, but let’s you get the traffic you do want. Those phrases will still end up showing up in your server logs, so that you can take all that for log offerings that people are typing, like how to eliminate Spyware. Find those log phrases in your server logs, and then add them into your campaign.
Eric Enge: Right, like when you have very, very general keywords that have potentially multiple meanings. For example, Jaguar, which could be a car, an animal or a guitar.
If you are selling the guitar, you know if you go broad match your conversion will be poor. You might want to use exact match on some 2 and 3 word phrases to keep yourself from showing up on animal or car related terms.
Rich Stokes: You can also negative match on the car and animal related terms.
Eric Enge: Right. Let’s talk about the cast of low coverage.
Rich Stokes: So, there is a common belief that if you bid more than anybody your ad will appear all the time. This is not true. There are people bidding on keywords where they are bidding higher than every last one of their competitors, and there ads are showing 5% of the time, and 95% of the traffic is on a table
Fewer than 3% of advertisers have greater than 20% coverage, so 97% of advertisers can increase their relevant traffic by five times by fixing their coverage problems. Coverage means that your ads are only showing up a certain percentage of the time for the terms that you are targeting. So if I type in Magnavox television 100 times, and it only shows up 50 times, than the coverage for that ad is 50%.
The goal here is to get your coverage to a 100%, you always want to be there, and there are a lot of reasons for that. If you have a brand, or something that could potentially be a brand, you want to appear every time that somebody searches on your keywords, because the brand equity that the impressions generate is very valuable. Secondly, there is also an effective brand recall when people click quit through, and they continue to see the same brand over and over again consistently.
It generates the perception that you are a company of importance, or brand of importance; whereas if you see a company once every 10 or 20 times, you might NOT consider them seriously. So, there is that perception as well. In addition, most people who have low coverage are generally over paying for the traffic they we get. The way I communicate this people is to say if your adverting budget for today is one dollar would you rather spend the entire dollar getting one visitor to your website of getting ten visitors to your website, meaning that they only cost you ten cents each.
You almost certainly get higher bid prices in cases with low coverage; and you’ll almost certainly benefit by lowering your bids that. Obviously, you get more bang for your buck; you are paying less for your ads, but also there tends to be less competition. You have much greater chance of getting higher coverage keeping all other factors equal such as ad copy.
Eric Enge: The idea is to manage your budget in a smart way to get the maximum amount of traffic for your investment.
Rich Stokes: Right, and maximizing your reach.
Eric Enge: Indeed. Alright, so what about a sweet spot for bidding; is there one?
Rich Stokes: Of course, there are sweet spots for bidding. So, we really used to report on this and go on at great length about how click through rates and cost per clicks change on Google AdWords depending on what position you are at. So, I think pretty much anybody on search advertising is comfortable with the idea that you are going to pay less with the bottom of the page, and your click through rate is going to be lower.
It’s probably one of the most well studied and well verified relationships in advertising. What was not previously done is a study to show the way keyword length factors in the equation. So, we figured that out, and then we broke that out according to the length of the keyword phrase. So, broad keywords; they have a different curve then there is for specific keywords. Broad keywords are well served at the bottom of the page; paying as little as possible. It’s typically only with the brand name advertisers who have deep pockets, and have a business model where they’re not worried so much about the direct sales, that can afford to be in the top positions.
Orbitz is a great example; every time you see an Orbitz ad, well they are cementing your perception of Orbitz in your minds, so the page should be up there. When it does come time for you to buy a trip, you become more likely to go to Orbitz or maybe one or the other two or three big name travel sites who have managed to build some top of mind awareness with you.
Now, to get into specifics; with the broad keyword what we found with our mathematical model is in most cases you are going to best served by having your ad in position 7. Any higher than that and the cost starts going up way up in proportion to your revenues, and so point of maximum profit in most cases is position 7.
Now, if the keyword lengths get longer you start getting at least 3, 4, 5 length phrases. You want to bring your bids up correspondingly, and so what we’ve found is that switch box for this next keyword is generally position 2. There is a real dramatic drop box in many cases between position 2 and position 1. So, in position 2 you might be widely profitable; position 1 you may be losing an arm and a leg. So, it’s real important we find the right position.
Eric Enge: It’s interesting that the whole thing is so dependent on keyword length, but when you have a one word keyword for example, everybody in that space knows about that keyword and they are going to bid on it. You are not to do any homework; it is probably one of the terms that the CEO would have mentioned to you if you asked him or her to describe the space you are in, and you asked him for a dozen words; it’s probably one of those words.
So, these terms inherently have brand value, and this is what drives the pricing up, and you certainly have people that are willing to lose money on the term. Even though in a direct response model they are losing money they are achieving their objectives.
Rich Stokes: That’s exactly right. For them, this may provide very, very cheap branding benefits. Even if you are buying those premium positions, the price is still going to be less expensive than a television campaign.
Eric Enge: Right. So, it would be interesting to see how that drives the pay per click industry in the future as more and more of those brands come in and do that bidding. So let’s talk about maximizing relevance.
Rich Stokes: We’ve picked keywords; we are working on getting those click through rates high, optimizing our bids and so on. So, you have three areas where you can maximize your relevance, #1 is your ad copy, #2 is the fit between the keyword phrase and your ad, maybe even your entire business value proposition, and #3 is actual landing page itself.
What you want to do is get those things aligned, and I think Google has done really everybody in the industry a great service by formalizing the idea that okay, these three things have to align with each other in some way. Not everybody likes it; and the algorithm is not perfect by any means, but I definitely think it was a smart move.
I think it should be obvious that if you are selling bikes you don’t want to do advertise on the keyword Mexico. So, let’s talk about ad copy. Ad copy is an amazing thing, one single character in an ad can change your click through rate by two times to three times very easily.
There are a number of these things, and we’ve included a number in the book on that. One great example is exclamation points. Google does not allow exclamation points in the title of your copy, but not a lot of people know that they will allow you to put one in line one or line two.
We’ve conducted a pretty large study of ad copy this year, and we found that a surprisingly high percentage of ads with high click through rates incorporate that exclamation point trick.
Eric Enge: Just make it look different.
Rich Stokes: Yes, exactly. There are a lot of different best practices that you can follow. Some of them we’ll be releasing out later in the year, but the idea is you want to continually click test your ads, find out which ads are the best. And, you have a couple of different models for informing you about how to do that.
So, the first tool is obviously the customer life cycle model; what keyword are you targeting? If you are targeting a browse keyword then don’t sit there and try to sell somebody something in your ad. It doesn’t make sense to put the price of your product in your ad, because very few people are going to be interested. Conversely, if you are targeting the Magnavox television; putting a call to action phrase like learn more might be mediocre at best.
Because, they maybe looking to learn more, but they might be looking for good deal. So, a better line there might be free shipping today only; that would be a huge winner. And, we see a lot of high impact ads with that call to action model appropriately targeted. So, the ad copy for testing is a real big thing.
The next thing is landing page optimization. This is absolutely huge, and very few people know how to do it properly. It’s part science; it’s part art. I have optimized landing pages for some of the largest companies in the world and have; I’ll give you an example of, I can’t tell you the company. But, within two months of landing page optimization we increased their online sales from 20,000,000 a year to 80, 000,000 a year, and that was all web. So, landing page optimization is very big thing.
Eric Enge: Excellent. The next thing really was Killer Ad Copy, and you’ve already started to talk about that a little bit. The process by which you invent your first few examples of ads and try them out, and enhance that over time until you get to the killer copy.
Fourth Speaker: We just released a tool. Basically what we did was we devised a way to figure out which ads were performing the best in any given keyword or even an entire industry. And so, we did that without having access to the individual advertisers click through rate data. It turned out to be so effective that we put it in a new report. So, anybody who is with the AdGooroo service will put their keywords in. Thirty days after they have their group setup, and we then have enough data, we can analyze the impression, behavior of the ads in question for each of these keywords.
We can tell them with a high ability of success, which ads are currently doing the best in their industry. So, that’s a really nice way to shortcut the whole learning thing.
Eric Enge: In addition to that of course it also makes sense to just keep testing and let Google do a lot of work for you, right?
Rich Stokes: Oh, absolutely. What we do is we start with the ad copy report. For example, when we started we were at 0.79% CTR and we increased it to over 1.5%, and that was virtually overnight. From there we kept testing, and we took the ad copy which was then the top ad copy, and then we turned it into something new by tweaking it.
We kept pushing the bar even higher, so not only do we stand out above the crowd, we get lower cost per clicks, higher click through rates; we often make it much, much harder for any competitors coming to our space in these edges.
Eric Enge: That’s great. Let’s wrap up with a beauty, Google Quality Score.
Rich Stokes: First of all there is a little bit of confusion about it. Google has two bots that come by and look at your site. One is the Googlebot, which indexes your site for web search. The other bot is the Ads bot, and that’s the bot that attempts to figure out how well aligned, how relevant your page is for the keyword in question.
So, each bot works by different rules. You know people try to gain the Google bots all the time, and they are succeeding less and less these days. The Ads bot however is a different story; it’s much simpler and I believe it works in a very similar manner as a Googlebot did in about 2002.
What we find with the Quality Score, Google is now assigning a Quality Score from 1 to 10; and the higher Quality Score the better the match with the ad text the better the match with the keyword, and the lower your minimum bid. So, if they decide that you’ve got a fantastic match, your minimum bid might be three cents or four cents. If they decide you are horrible match, your minimum bid might be $10. Very few people can afford to pay $10 a click for anything. And, even if you could; your coverage would be so low that your ads will hardly ever show.
So, it makes very little sense to try to use a brute force approach to the Google Quality Score. Rather what you have to do is you have to investigate your page; figure out what are the factors that are causing you to have a low quality score. What we found in our research was that many of these factors not all of them, many of them were onsite factors, things that were 100% within your control.
Having the keywords in question on your page is a factor. It’s not a necessary factor, as the overall genre of your site may be consistent with the phrase they are targeting. But, most of the time it makes a lot of sense to have that word, that phrase on your page. Another very big thing; this is one thing that they have tweaked over the last six months is the page loading time.
What we found were that pages written in certain technologies, tended to have lower quality scores. This happens in technologies that insert large binary strings in the code, or otherwise bloat the page code. This slows the page load time and harms the quality score.
Eric Enge: Page load time is certainly a very interesting dimension, but you can see how it relates to the notion of quality. I mean it takes 5 seconds for the page to load, and the competitive page takes one, well what’s a better user experience?
Rich Stokes: Right, exactly.
Eric Enge: So, there is something that really deserves some real consideration.
Rich Stokes: The great thing about this sure is that you can test this in real time. It’s something a lot of people don’t realize. Google has recently said that they are going to be updating page of Quality Scores a lot faster if not real time. I don’t know if that’s true or not, but it was three months or four months ago.
In the past, you had to wait three weeks to get an update on your Quality Score. So, we thought of a way around that. If you want to test two different landing pages, what you do is create two different ad groups. These ad groups are identical; they have the same keywords, same matching, everything is the same. The only difference is you give them different landing pages.
You’re doing an A/B Split test. Then, what you do is you take the handy-dandy Google AdWords editor which is a Desktop Application that they give away for free on the site. And then, you load up your campaign in that. That gives you a way to look at all of the minimum bids required for each one of your keywords.
Let’s say you’ve got 100 or 150 keywords in your group that you are testing. You may find that five of them have a $10 minimum bid; maybe 30 of them have a $5 minimum bid. Ten have got a $1 minimum bid and so on; you can group those up and you could see how many keywords got cheaper, and how many more keywords got more expensive with both versions of your page.
Using that you can very, very quickly figure out how to optimize your page to get the lowest minimum average bid on Google. You can do this in less than a couple of hours which provides an incredible return.
Eric Enge: Thanks Rich!
Rich Stokes: Thank you! The pleasure is always mine.