John Marshall
The following is the transcript of an interview of John Marshall, CTO of Market Motive. John was CEO of Clicktracks, until very recently when he left the company to put together Market Motive. A Netscape alumnus who is responsible for two other successful software companies, John Marshall brings 20 years of experience in the software industry to Market Motive.
Interview Transcript
Eric Enge: In a recent interview you did with Ralph Wilson, you spoke about buying traffic and buying intent. Let’s talk about that a little bit.
John Marshall: Yeah. I think; I guess I should start out by explaining that I am of the opinion that all visitors to a website are not equal. When you use the word traffic, it describes visitors to your sites. You are putting everybody into the category of people that drive by the site and have only looked at the site by chance.
If you view your visitors as traffic, then you view them with indifference. I strongly believe that people come to a website, even a small niche eCommerce site, with varying degrees of intent. Those people that have very low intent, you might describe as traffic. But, that’s not a reason to use that word to describe everybody. I want people to think about the intent the people have when they come to their site.
If you use the example of search marketing, that whole channel has been so successful because when you do a keyword search, you express so much of your intent to the search engine, and the search engine gives you results and you click through to a site. The sites are successful by capturing that intent. That does not mean that people clicking on links to your site from other sites have anywhere near the same degree of intent.
They are frankly highly likely to have a lot less intent; in some very rare cases, they might have more. So, if you use the word traffic to describe people coming from search engines, and you use the word traffic to describe people coming from spam, like a spam email campaign, I think that’s a mistake. You shouldn’t view those two very distinct behaviors in anyway near the same category. They are very different; one has strong intent and therefore is very valuable, and one has very weak intent and is therefore almost valueless in my opinion.
Eric Enge: Right. Now, I don’t know valueless is the same conclusion that I would jump to, but I would agree that when you have someone who has come from a search engine; they had to do some work to get there and expanding upon a scene that you have talked about lately, I refer to this as they have made an investment of time before they get to you.
Whereas, if they are on another website and they see an ad there and they happen to click on it, well they didn’t really make an investment of time to go to the search engine and then enter a query. They really had done nothing when they just clicked on a link, and it’s not that far removed from TV advertising, which is interrupt driven and essentially invasive in nature. It’s not quite the same as that of course; perhaps that’s even further out on the spectrum.
John Marshall: Yes, indeed. I don’t like the idea of putting all of these different types of behavior in one bucket and referring to it as traffic. Actually, I’ll give you a contradictory point of view to mine. My point of view is people who went to a search engine have strong intent, and people that get to you via banner ad have weak intent. Now, one of the things that we’ve recently looked at is direct navigation.
For the benefit of the reader, this is where people type in a URL, which they think relates to the topic that they want to research. So, let’s say that you want to buy a new car; some people would go to a search engine and type in new car pricing. But, some people would go to the URL bar and just type in newcar.com
I think this comes about because domain names tend to reflect the business of the company which is often the domain name. So, it’s not unreasonable to do that. Personally, I wouldn’t do it, but thus there is also the fact that people do. Based on my statement now about traffic; let’s imagine that you are an advertiser. You can actually buy these clicks; you can because the company that owns the domain name newcar.com will, in fact, lease the clicks to you.
So, let’s say that you are a car retailer; they will lease those clicks to you so that you get the traffic. Here I am using the word probably appropriately; you get the traffic for this activity. Based on my statement about intent, I am sure you would agree that that traffic is going to be poor quality. It has weak intent because it’s a different model; it’s a different behavior than going to Google and doing research, but there is less investment of time and effort.
However, I can tell you that actually, that type of stuff can work quite well for some people, particularly for more inexpensive products. Those types of campaigns, in a theoretical sense, kind of make me clench. But, they actually work; we have seen it work quite well. So, there is a contradictory point of view to my own idea.
Eric Enge: Well, let me contradict your contradiction and say, but if somebody goes and types in www.newcar.com; they are once again taking an action and displaying intent. They constructed a name in the search for a site that sells new cars. So, I think not really very different than going to a search engine and typing “new car”. That would be my take on it. They are formulating their query, and instead of putting it in a search engine they put in the address bar.
John Marshall: I think that there is truth in what you say; that some of that carries good intent. But, I am still surprised that sometimes these direct navigation behaviors actually yield reasonable conversion. Of course, it may depend on the particular type of universal keyword you are targeting.
But, I am known as saying that this type of stuff has weak conversion characteristics. I have been surprised to find a couple of examples that have good conversion characteristics. What I always boil down to is does the user have strong intent. If they don’t, you get poor conversion characteristics; even though you may get lots of traffic.
So, this traffic thing; there is just something about it that I don’t like. And, it’s the idea that traffic means stuff that is drifting past in front of you. We look at traffic on the road, or we see a lot of vehicle, lots of traffic. We don’t really understand where that traffic is going, or what it’s doing.
If we are headed to a beach, when we get to the destination, it’s no longer traffic, right? People actually doing stuff, so traffic is really the role metaphor to use for people that are coming to visit your website. And, I don’t like the idea that people with a web business think of human beings coming to their site in that way.
Eric Enge: Right. Basically, they are leaving a lot of opportunity on a table when they do that. And so let’s take the concept a little further. So far we’ve talked about direct navigation, and banner ads, and emails, and search engines. But, when you have something coming in from a search engine you can detect bits of intent by looking at the keywords, right?
John Marshall: Right. We also need to consider the search results page that they came in on. Somebody who clicks into your listing on page 1 of the search engine results has a certain amount of intent. Somebody that clicks in from page 20 of the search engine result probably has even stronger intent. The irony is that you don’t necessarily want to be on page 20 of the search engine result, but, if you are on page 20, those few people that you get have really strong intent.
Eric Enge: Yes, although some of those people would just be people researching competitive sites, right?
John Marshall: Yes. That’s true actually.
Eric Enge: But, I do understand the point, because they had to work a lot harder and they aren’t satisfied for whatever reason in the earlier pages that they looked at, and hopefully the experience would be different when they get to you.
John Marshall: Yes, exactly. So, from all these different sources, people have a very different intent, and a good website should be able to adapt to those varying degrees of intent.
Eric Enge: Right. Of course, sometimes that kind of activity requires a dynamic web application if you are going to be interpreting that kind of data.
John Marshall: Interesting question; I am not sure that I would assume that. I think you can build a very good website that adapts and deals with these different degrees of intent completely with static pages.
You don’t necessarily need a dynamic marketing suite or one of these complex systems, and I am skeptical those things really work. I mean personalization actually works. I think for many companies it’s overly complicated. Well written copy that adapts to a particular situation under which somebody is coming to the site, I think that is the big lever that really moves the results.
Eric Enge: Right. So, when you talk about marketing copy on static pages, you are really talking about recognizing the kinds of intents that people have when they come to your site. And, allowing them using their own eyes to find on the page where they get the task they are looking for.
John Marshall: Yeah, exactly. So, an email landing page is going to be very different than a PPC landing page.
Eric Enge: Sure. But a different issue is discerning what to do using an analysis of the search keywords used. For example, you have someone who searches on “buy new cars” versus “new car reviews”, yet the search engine has decided it’s the same landing page in both cases in the organic search results.
Then you either have to cover both scenarios on your page in a very clear way for the user in a static application or a dynamic web application which presents a bit of a different experience.
John Marshall: Yes, indeed. And, in some of those cases, you might need a dynamic system or at least a simple mechanism whereby you can redirect people to an appropriate page. But actually, yeah that’s a very good example where the intent is different, is pretty clearly expressed in the keywords; and you should be able to reference that.
The probability that you can get a search engine ranking on the same page for search phrases that have a very different intent, I think it’s quite small, right? I mean it’s possible that your page contains multiple meanings and different keywords could come into that page, but you are unlikely to get a listing and a ranking for very different meanings of a page
The search engine has got to ascribe a certain meaning for the page not against the keywords. And, it’s only going to do that for a subset of keywords that mean substantially the same thing. So, therefore my opinion is that a well crafted static size is most of the time good enough to be able to tease apart all of the intent that you need to.
Eric Enge: Right. Even if we go to the world of organic search, which is different; and let’s say if you use digital cameras as an example. Your page for buy digital camera is probably different than your page for digital camera reviews.
John Marshall: Exactly. Even if you’ve got one page that could address both of those things, most likely you are going to rank for one or the other. The search engine has got to ascribe a certain meaning to the page, and if you want to rank for both of those terms, you are probably going to be forced to build two separate pages and have anchor text linking into those two separate pages with two separate anchored keywords from other sources.
Because end pages need to have a single meaning most of the time, if you think about it ahead of time, you can build sites, and pages, and campaigns, and keyword buys that do correctly address the different intent that users have. The mistake is in thinking that everybody has the same intent. Once you realize that people have a very different intent, it’s pretty straightforward to address, you’re your biggest challenge is writing copy and trying to decide if you should be using long-form copy, or short form copy, or whatever.
Eric Enge: Right. To expand on this, we have the notion of researchers versus shoppers. I’m sure that many other types of people who come to the site, but those are two major ones that come to my mind.
John Marshall: We’re restricting that to the context of an eCommerce site, right?
Eric Enge: Yes.
John Marshall: Yeah. I mean, yeah I think that’s right. There are researchers, shoppers; I mean it really comes down to where you are in the buying process. The set of keywords that you buy to capture people at the beginning of a buying process, are very different than those that you buy at the end of a buying process. At the beginning, it will be things like digital SLRs, or digital cameras, and at the end, it will be specific products and model numbers, and in between reviews. So let’s use a couple of examples; what’s something that you’ve brought recently online?
Eric Enge: I bought some organic food online.
John Marshall: Organic food; actually that’s some very good example. What were the keywords that you used; maybe we are getting too personal here, but it’s kind of an interesting example.
Eric Enge: I could get into some very funny answers to that question actually. Seriously though, I started by looking for a green nutrition bar or something like that. So, that might have been the initial search, “green nutrition bar”.
John Marshall: Okay. That’s a really good example. Did you mean green as in the color or as in the ecological movement?
Eric Enge: Well, I meant it in the sense of eating greens at dinner. You are supposed to eat leafy green vegetables, and you are supposed to eat beans. So, it’s the color, but there are actually some additional connotations to it when you apply it to food.
John Marshall: Interesting. At the beginning there, you were doing research, right?
Eric Enge: Yes, I was trying to find out what the right thing to get was.
John Marshall: Did you do a separate search then, when you went to finally buy the product. In other words, did you use one set of keywords at the beginning and another set of keywords just when you make the purchase?
Eric Enge: Yes. As I got closer to the purchase; I started to have some product names in mind, so I started searching on those.
John Marshall: I think that’s a very good example then. Your intent increased as you went along. I feel pretty strongly that in these days most things are bought in this way. I think maybe it’s important for people to understand that that’s something which has changed. 3 years or 4 years ago buying something online; the choices that you had and the difference between those choices were much smaller. So, you tended to not have these two separate phases of research, and then purchase, right?
That tended to not exist because it wasn’t worth doing them separately; because the difference that you would find was very small between different products and different sites. Now, that pretty much everything has moved online, there is a big difference between those two phases.
I think some marketers haven’t adapted to that; they are still stuck in this mode of, it all happens in one go, it all happens at once. And, they really fail to capitalize on the difference in the behavior between the research phase and the purchase phase.
Eric Enge: Right. It is a very different set of circumstances, and part of it is realizing that the majority of purchases involve multiple visits.
John Marshall: That’s right. Most web analytics tools don’t properly reflect the multiple visits to a site. Some reflect one visit; some reflect only the first visit, some reflect only the last, and there are completely only a small number of tools that correctly give you data for two distinct visits.
Eric Enge: Yes. And, very few will allow you to configure how you want to weight the allocation of revenue to each visit.
John Marshall: Yeah. Are you a fan of that Eric; do you think that that’s something that should be configurable; a configurable weighting of revenue attributed back to the keywords?
Eric Enge: It just seems to me that how you want to structure that has got to be dependent on the nature of your business, right? I mean the classic case for me is you have some system, and it allocates the revenue based on the last visit, right? And then, a person goes back to their bid management tool, and it’s telling them to disable a keyword that is often the first keyword in a user’s buying sequence.
For example, a phrase like “digital camera”, it is telling them to disable digital camera, because they are not making money on it. And then, they are cutting digital camera off and they stop bidding on it; and their margin goes down. And, the reason why it goes down is digital camera was driving a follow on visit to find digital camera reviews and finally to a model number. And, a good percentage of those model number visits occurred only because of the original searches on digital camera.
Eric Enge: If you allocated all the revenue to the last keyword, your digital camera keyword is being shot changed. That’s the classic problem, and, it’s leading into taking incorrect action.
John Marshall: Yeas, I agree. I mean I think that you’ve got to be able to have both. I just feel that configuring the attribution between the past and the most recent keyword is a level of complexity that a lot of people are not willing to take up.
Eric Enge: I agree. Even if they are willing to take it on, what are the right numbers to allocate throughout the process? The challenge is that most bid management tools are structured to take that data and inherently use it the wrong way. They use it in a convert it on this transaction or get downgraded approach.
John Marshall: Yeah, interesting. So, where does that leave the poor advertiser? Well, I guess the only thing I want to say about that is the difference in intent between the research phase and the buying phase then, or early stage and late stage. As long as you grasp that, then I think you’ll be okay.
Eric Enge: Yes. You have to be knowledgeable about the weaknesses of the things you are working with, and of course, understand their strengths. And, apply a little bit of a filter to what you do, based on the knowledge you have.
So, let’s talk a bit about persuasion, and I guess that’s the other part of this now in my mind. You’ve determined intent, and now you are trying to get a conversion if not in this visit, in the future visit; let’s talk about that a little bit.
John Marshall: Yeah. I think this is another example of not all people coming to your site are the same. Some people coming to your website want to make a decision quickly, and to paraphrase Bryan Eisenberg, they tend to be competitive type people; they want to make a decision quickly and move on. But, some people are very methodical; and some people are rational, and some people are emotional. So, you’ve got this spectrum of different people coming to the site.
Some products and some sites tend to bias in a certain direction. So, our website selling industrial widgets is certainly going to appeal more to methodical people than it is to spontaneous people; because most engineers tend to be methodical and not spontaneous. But, nevertheless there is some degree of spontaneity in some people who are coming to the site for industrial widgets, and conversely, a website selling fashion accessories is going to appeal more to people who are emotional and spontaneous than it is to people who are methodical and rational.
So, you’ve got all these different personality traits if you like; and also they come in different degrees at different stages in the buying cycle. Research versus actually making a decision to buy; research tends to be done by people who are methodical. So, you also have somebody who appears to be spontaneous; they may certainly become methodical when they are faced with the checkout page, where they are going to check all of the details there if they are unsure about that thing.
So, the great challenge is in building copy and a structure that appeals to all of these people. In the end, you almost can’t appeal to everybody equally. You’ve got to choose who you want to appeal to, and also you can structure the site to channel people in the right direction so that they find the content, which is going to appeal to them most. You can do that by structuring the copy in the site so that methodical people are going to tend to click on this, and spontaneous people are going to tend to click on something else.
Once you’ve got that initial starting point set, you can actually build different copy, which is going to appeal to these different groups and that’s going to allow you to persuade them. Interestingly enough I said earlier that a static website would be good enough for doing a lot of the necessary, intent modeling that people need to do. In many ways a static website is better than a dynamic website for this kind of stuff because a static website allows you very carefully craft copy and you are not forced into a template based system, that requires you to make all pages the same, alright?
A static file I can convert very, very well for this kind of thing. Typically, people build a single website those appeals to one type of person, because their content management system railroads them into that method of thinking, and that’s a mistake. Not everybody comes to your website for the same reason, and you might need a quite different look and feel on the pages to appeal to the broadest possible audience.
Eric Enge: Right. So, when you are talking about persuasion of people that came in from a search engine on the same search term and you are trying to segregate them into groups as well, right?
John Marshall: Yes, correct.
Eric Enge: And then, it’s important for people to understand that each person can go through all four stages. You certainly have people that are very competitive people, and who are more often in competitive mode than others. That doesn’t mean that they don’t have times when they are in a methodical mode or emotional mode or whatever, which they fall into in one moment or another along the way.
John Marshall: Exactly. The question then is how you would factor that in to your own website, and that, of course, is the whole model of persuasion. In many cases, the trick is really quite simple. It all hinges on writing some copy that allows people to self-select, where a methodical person is more likely to click into page A, and a competitive person is more likely to click into page B.
By giving people the choice, you can determine the stage they are at. Based on that, you can then present them with the copy which addresses whatever concern they have at this stage in the buying process for that particular personality type.
Eric Enge: Right. When you talk about self-selection, you are talking about offering them a really clear menu, or a system that allows them to find their way into their happy spot. Then, the next thing is the notion of building confidence during the process.
John Marshall: Right. How do you do that? Well, that’s the art of being a copywriter; and the world needs more of them.
Eric Enge: Yes. You have to start putting some thought into when someone is in competitive mode what they are going to want to see to make them confident that it’s the right decision to their investigation or make a purchase on your site.
John Marshall: Yes, exactly. For example, a competitive person wants something that is better than what other people have. So, for a competitive person you would want to emphasize the product that you are looking at is better because of this, this and this.
Presenting that type of argument to a methodical person could be a mistake because a methodical person doesn’t necessarily want this to be better than everyone else. A methodical person may want to know that it is backed by a guarantee, and that it has these certain features which make it long lasting. The copy which appeals to person A might actually repel person B.
In many ways, that’s why a lot of sites don’t implement this stuff. They don’t implement it, because the person writing the copy knows that writing copy that appeals to both the person A and person B is very difficult.
They haven’t built and structured the site so that these people self-select and they view the appropriate content. The content itself cannot be made to appeal to everybody because, of the fact that it repels certain people. You actually can’t take this persuasion concept and bolt it on top of an existing site. You really need to design it into the structure of the site from the beginning, and that makes it a difficult problem for many people.
Eric Enge: Right. Let’s talk about another thing that I think really ties into the wide wonderful world of conversion. For example, is it better to simplify a page or pack it with content? Any specific experiences you might be able to share about that?
John Marshall: Yes. Instinctively, I would always say that the less copy works better.
Eric Enge: Less is more?
John Marshall: That’s right. I like things to be simple. In most cases that we’ve seen long form pages work better, it is because the copy contains enough information somewhere in there that will appeal to everybody. But, it’s not necessarily true that everybody should go and build lots of copy into their pages. In addition, you also need to think about whether there are lots of navigation points or too few navigation points? The amount of copy and the number of navigation points often are proportional to each other.
I think that fewer navigation points work better, and lots of copy works better. But, those two are quite difficult to achieve simultaneously. And, when you’ve got lots of copy, it tends to be because you’ve got lots of navigation points. For example, you’ve got a huge left menu; a column format left menu with all kinds of different products that would be very common in an eCommerce site.
Because, you’ve got all that junk down the left-hand side, there is a temptation to fill the page therefore with lots of copy; because you’ve already got the space.
But, you may well be presenting the user with too many product choices. But, in an eCommerce site, you can’t always control this.
Eric Enge: Right. In that environment don’t you have a situation where it’s almost like the repeated menu sections and things like that just don’t count, because they just get blind to it.
John Marshall: I understand the point, but in general less navigation works better.
Eric Enge: Interesting. So, what about other thoughts about improving site experience or persuasiveness that people can take away as this is something I ought to think about.
John Marshall: One of the best things that you can do is to get somebody to audit your site. It brings huge benefits. So, let’s imagine that you’ve got a certain amount of money to spend on improving your site and let’s say that an audit of your site costs to a $1,000, or $2,000. That can yield an enormous improvement and there are all kinds of different people who do it. We have done audits for people, and the results are always very important to the business, and it brings about a huge change.
Up until recently, we didn’t do audits for people. But, since we started doing it we have really seen the benefits and, I have become a bit of an evangelist for doing it.
You can also just ask people or do a survey. Run a quick little survey for people that arrive at the site, why you are here; did you find what you are looking for, what can we do to improve the site, etc. iPerceptions has a free tool that will help you do this. These types of tools are very powerful can reveal much more about the intent of your visitors than you can discover with web analytics.
Eric Enge: Yes. Your partner Avinash has often suggested that tools like that are really a part of web analytics.
John Marshall: Yes. When I was involved only in web analytics, that’s how I viewed the world. But, once we started to move beyond that, we really began to understand that we did not correctly understand the intent of users simply by looking at the numbers. You have to get out there and ask them, and a survey or an audit is a great way to do that.
Eric Enge: I think some people are afraid of angering their users, but the iPerceptions solution, for example, strives for minimal intrusiveness and makes it a lot easier and more approachable for people to try.
Let’s talk a little bit about some of the things going on in the world of bid management. Interestingly, Omniture has made a big push in this are of late. I spoke recently to Gail Ennis, their chief marketing officer about it; and evidently, they have something like $600 million under management at this point.
John Marshall: One thing I would say is that I think the word bid management has become a misnomer. It used to be that a major pain point that people had was in adjusting their bid prices. I think this was largely because of the strange way that Overture did their bid pricing
Therefore the product space became known as bid management. I am of the opinion that adjusting bid prices is a lot less of a problem now because the price is no longer nearly so directly controllable by the advertiser. Because of things like Quality Score, it’s actually quite difficult for the advertiser to adjust the actual price. They have got indirect control over that at most, and I think therefore the term bid management doesn’t reflect the problem that the advertiser has got.
The problem the advertiser has these days is ad management. They’ve got lots and lots of ads to manage; they have got tracking URLs with just complete anarchy going on. They struggle to come up with a reasonable taxonomy for the campaigns they are running and a hierarchy for them. If they’ve gotten more than a thousand keywords that they are buying, things very quickly get out of control.
They struggle to know when they are running split tests on ads, and which ads have still got a split test running.
That’s where these tools can make a big difference. Of course, they all have a natural fit with web analytics because that’s where you get your data. But, I want to encourage people to think of them less in terms of bid management and more in terms of ad management.
Eric Enge: Right. Do you see expansions in ad management capability as being a major investment for the analytics vendors then?
John Marshall: Yes, I do. It is such a huge pain point for users. And, you could argue that the search engines are going to constantly improve the user interface, and they are going to make that process better. But, when everybody is advertising across three different platforms, it just becomes unmanageable. So, you need a tool to help you with that.
Eric Enge: Right. We talked a little bit earlier about some of the issues with multi-session purchases and in fact a little bit about how that affects bid management. But, you can also extend that a little further and talk about cross channel marketing; so a pay per click ad that results in an offline transaction and things like that. Do you see any developments in areas like that coming soon?
John Marshall: There is stuff definitely happening. I think that many companies have a problem with implementing cross channel marketing measurement. For example, the company may have a CRM system, which is tracking the offline sales process.
The belief for many people is that online activity should populate into the CRM system since the CRM system is where the salespeople spend their time; the data should go into that. But, the reality is they work the other way; they work by taking the offline data and populating it back over into the online system.
Bid management companies struggle with that, and I don’t like doing that because it potentially exposing tentative information about the customer, and their physical address and various things like that into this web analytics system. Companies don’t like it, but the hard reality is that’s probably the way it’s going to work. Because doing it that way lets you work with a smaller amount of data which you can easily transfer.
Try and do it the other way around, we are taking your online data and populating that into the sales force system; there is just way too much data in it, and it can never work. I think the reason that you are not seeing very much integration there at online, offline is the technical implementation requirements are ugly, and people haven’t got used to it yet.
Eric Enge: Thanks, I enjoyed it!
John Marshall: Thanks, Eric!
