Keyword selection for the uninitiated

This week I read an article by Sandy Parish about the importance of keyword selection strategy. It posed the question that new-to-search-marketing people might ask, which is: Do I use specific terms or do I use broad terms. She elaborated on the virtues of modified broad terms. I agree with her point, that if you are considering broad terms then modify them to be more relevant to your specific offering.

For instance, say you are selling a book and it happened to be a police drama. The broad term for search would be “books.” The modified broad terms would be “police drama books” or “drama books” or “literature drama”. You get the idea…there are many ways to go. You use the modified terms, arrange your ad copy accordingly, and your relevance goes up, your cost goes down and or positioning improves.

In addition to the modified broad match approach, we can offer those new to online marketing (in particular, search), even more guidance. The underlying premise being:

Start with the basics, the obvious; limit your financial exposure and move on from there.

For those of us who have been in the game for a while the term “basics” has a broader scope, and greater financial exposure is less risky, being mitigated by our experience. For the truly new, consider the following.

Keyword selection
Start with the narrow, but don’t ignore the broad (I’ll explain later). Use company name, brand, product, style, model, etc. For example, you’re a retail store selling shirts. Pick your top sellers and start a list. Think both in terms of how the products are marketed and how consumers talk about them. The apparel industry talks about “dress” shirts, but if consumers refers to them as ‘long sleeved’ or ‘business’ shirts, then use these terms as well. Your list might look something like this (abridged):

Company:    My Shirt Shop
Product:    Dress Shirt
Brands:     Boss, Faconnable, Ike Behar.
Styles:     Classic Dress, Tailored Dress,
Patterns:   Solid, Stripe, checked,
Colors:     blue, red, white, pink…

Obviously, your attributes and ultimate list will depend on the product or service you’re marketing.

Then, mix and match in a way that is logical to your industry and products. This could get quite long. The engines have keyword suggestion tools that you can use. However, when it comes to developing a comprehensive list, you know your business, don’t underestimate yourself.

Examples of mix and match:
Boss Classic Dress Shirt
Faconnable tailored dress shirt
Ike Behar blue solid tailored dress shirt
My Shirt Shop Shirt

As I mentioned, this can get long, that is why you start by limiting the scope to your best sellers (remember, this is for those new to search).

Ad Copy
Each of these keywords should be set in like groups so that ad copy can be tailored to them. For instance all keywords / phrases with “Boss” & “Blue” in one ad group, so the ad copy can include these terms and be relevant to all keywords. How far you take this is up to you. Technically, the farther you take it, the more relevant your ads will be and the better your results will be. Also, the destination URL should be to a page that is as close to the keyword / ad copy as possible. Don’t send them to a “shirt” page if you have a “Boss Shirt” page. Relevancy is the key to efficiency in search.

Now the question comes in, what about the keyword “shirt” or “dress shirt” as the modified broad. Here is my advice. Use them. But, limit your financial exposure. Create separate campaigns for these keywords, and cap the daily spend to an amount you are willing to invest in learning. With a CPC of $0.50, capping your daily spend at $10 gives you 20 clicks a day, 600 in a month with a spend of $300. During this time, you can adjust ad copy for relevance to improve results and move the CPC. Before the month is out, you should know if the conversions at $0.50 (or any other amount) are enough to justify the cost. These numbers are simply for demonstration, but the principle applies at $0.10 or $2.50 CPC.

In reality, this same principle applies to the entire program. Segment it, cap the campaigns at reasonable amounts and adjust the CPC and copy to make the program profitable on a per click basis. Then, when you have the CPC range set, lift your daily cap.

Now, expand to new keywords, monitor, adjust, expand again, monitor, adjust….you get it.

One of the quandaries posed by Sandy was that a too limited keyword set would not drive enough traffic. I would pose that the long term view should be taken. Set a foundation of good practices that can be scaled and the volume will come.

 

Ironic that Mobile search is paving a cow path…

The mobile market represents the next wave in search. However, I see two current problems: 1) the current scope of ‘innovation’ is geared to subject matter that consumers currently seek (ring tones, images), rather than looking to the next level and 2) new search methods are mostly slight improvements on the difficult keying and navigation process. Even with this there are some neat things going on.

Microsoft purchased MotionBridge to enhance the mobile search capabilities for Live.com. It does appear to be a better way to do things. But, it also appears to be the proverbial paving the cow path. Rather than the typical list of links, you enter the phrase in a single field and MotionBridge returns a category segmented list of links. In this example you can see that, depending on what you type in, there may be a predefine list of links from the portal, or search results segmented into types of content. They also provide numeric codes for certain topics. For instance Sports is 77678 (I don’t know that I could remember the code to enter for my interest). The limitations here continue to be the reliance on the keypad as the input device. Because this is the LCD for most devices, it makes sense. But, real innovation challenges “what makes sense” and offers us a new and better way.

Medio takes a different approach by trying to predict what it is you actually want, and giving you the information rather than links. The example shows that they believe most searches go uncompleted because the trail you need to follow is too long. Through personalization, Medio is betting that they will know you well enough to present you with the answer to your query, foregoing the need for a list of links. I have to agree that, given the limited scope of subject matter that people seek on mobile search, this predictive modeling will be okay. But again, text input is a limitation. Also, the algorithms are based (necessarily ) on how people currently use the service. Those who use search with any regularity know that the ability to guess what we want is not terribly accurate (how often is the first returned result where you stop?).

Late last year, Medio and Verizon partnered to provide VZ customers easy access to ringtones, games and pictures. This is the equivalent of vertical search online and it makes sense given that the Medio's purpose is to deliver the answer to the query rather than links to possible answers. It replaces knowledge of the user with a well defined scope. As our interest in mobile search subject matter expands, this becomes less possible.

One of the more interesting applications is the image recognition query. Snap a picture of “something”, send it SMS and you will receive related links. Mobot has built an application that does this. Imagine having an ad in a hotel tourist book / magazine for your restaurants, with a message saying take a picture and send it sms for a return of coupons, maps, reservation function, etc (until the functionality becomes widely known, the message will be needed) . Suddenly the magazine ad becomes an easy way to share more information and facility business. If you have a recognizable logo, you can have consumers send in the picture and have any number of things returned; ring tones, locations near the person, promotions, WAP links, etc. This application, though limited right now, provides a view into one possibility of initiating a search query (I know that sending a short code via sms can do similar things, but the pattern recognition is the key point here).

It was interesting reading Gord Hotchkiss’ epiphany  article, "I have seen the future."  He wrote about the next generation of processor chips brining us closer to the possibility of voice recognition. This, when it happens, will be a gateway into real innovation.

A case for your value as a marketer.

I read Aaron Goldmans post, “Should Marketers Outsource Search" on Media Post, and a flood of memories came in.

Before I get into my reaction, a little background will lend context. I started off in DM with direct mail and telesales for a local theatre. After a couple of years, I joined a large ad agency in the media group becoming a planner and buyer for television, radio, print, and OOH for consumer package goods and automotive clients. My next stint was in a national yellow pages agency. By 1994 I was working with clients and account teams to develop YP strategies as well as developing our first on line ads programs (can you say Prodigy? – if you have to ask, don’t bother), which naturally morphed into SEM. Anyway, I came to the YP clients with brand experience, strategy development and segmentation know-how derived from a fairly diverse background.

I can tell you that most of the clients we talked to thought Yellow Pages was a no brain-er. The only thing that kept in-house departments at bay was the fact that it was labor intensive. You couldn’t just bring in a person and have them do it. So, how is this relevant to SEM outsourcing? What it did was provide the opportunity to demonstrate the unique skill sets and abilities that are really required for the “no brain-er” media. Eventually respect for our value was achieved by:

1) Learning and clearly demonstrating our knowledge of the clients’ industry, brand and products. Professionally, this put us (if not yet our medium) on par and in some cases above the client in their eyes (you’d be amaze by how many people know their brand and products, but lack insight into the industry as a whole).

2) The next key factor was using this knowledge to develop media specific strategies that tied in very closely with the general media and overall company strategies. It was through the strategic use of our medium that we demonstrated to the clients that not taking it in-house was a wise strategic move as much as it was a good financial move.

Lessons learned: Don’t allow yourself to be relegated to talking only about search. Engage the client in strategic applications beyond search and beyond online. Through a consistent application of your knowledge beyond the basics, beyond your medium, you instill a level of respect that earns you a seat at the media table. At the large agency, we always started with the clients’ business objectives, then tailored the media strategies accordingly. There was no presumption on the part of the client that they could take media buying in-house. Not because buying a :30 second spot was too difficult. But because they did not have the knowledge needed to develop the strategies to know if a :30 second spot was even the right answer (actually, we all knew it was before we asked the question ).

I have since left the yellow pages industry. Interactive pretty much consumed most of my time there, and for a while now, consumes all of it (professionally). But the same lessons apply. I see our client services team demonstrate these realities every day. I can not remember the last “search” conversation I had with a client that did not expand well beyond search and into their business.

Share matters, but campaign performance matters more

When it comes to talking about the search engines, we inevitably get to (if not immediately get to) market share. There are three sources most frequently quoted: Nielsen/NetRatings, comScore and Hitwise. A lot of attention is paid to these guys when they release the numbers. Personally, I do not follow these numbers with great rigor. Rather, I prefer to follow the performance of our individual search programs which span the more and lesser used engines. If there is a major share shift (unlikely), I’ll see it before the reports come out.

That said, there is a long term view that the ratings can provide. I think following Danny Sullivan’s perspective is good (essentially, not reading anything into changes that do not consistently breakthrough share bands). If you follow the measurements, it helps to have a perspective on how they get their numbers. This is not a statistical description (I’m not qualified); just a top line. For a good review take a look are Matt Belkin’s blog. He covers pros and cons of the panel vs click stream data methodologies.

Nielsen/ NetRatings  utilizes a panel method (computer resident tracking)(NetView) and site-side technology (StieCensus) launched in September 2005 (Integrated to overcome the cookie deletion issues).

comScore’s uses a panel method via qSearch with proxy technology tracking the users web behavior.

Hitwise  uses ISP data to anonymously collect click stream data.

One thing to keep in mind is that these players are also tying site traffic analytics into off line data – shopping, behavior, demographics, lifestyles, etc. So, search market share is really a small part of what they do. It just tends to get the most attention.

If you follow the engines (U.S.), you know that Google is huge (48 -60%), Yahoo! is big (25-30%), Live is small (9-12%), Ask and AOL are smaller (about 5%) and the rest are tiny.

Okay, I didn’t throw out the actual numbers. That’s because the true measure of the search engines’ strength is how they perform for your program. It’s the relative ROI that is important to you. So, keep an eye for any major changes in the engine share, but don’t ever lose site of your results. (SeachEngineWatch is a good place for the numbers.)

Look at some of the smaller players. There is a difference between share and efficiency. I have seen some small players come in with decent ROI (better than the big three). While they are not individually scalable, in aggregate, they can add net program efficiency.

For example, if a keyword on Google has a CPC of $0.90 and converts at 25%, your cost per conversion is $3.60. If, on MIVA, your conversion is only 5%, but your CPC is $0.10, then your cost per conversion is $2.00. Now, Google may get you 10,000 conversions and the aggregate of the others may only be 100, but at a 45% discount, these can be very efficient.

The key here is measurement. If you can not track to the conversion, these sites can just as easily be money losers. If they only convert at 1%, then your cost per conversion jumps to $10. While this is true with the big engines, most search marketers pay attention to them. The big ones are the fire hoses that we turn off right away vs the little ones which are tiny drips that we let go too long. Good metrics helps prevent this mistake on either end of the spectrum.

Until now, I have not touched on click fraud. I don’t discount it by any means. But, to me it is a different, though related, discussion. If you have your metrics in place, even if there is click fraud, you’ll be able to mitigate it, or even profit in spite of it. It is difficult to continue a program where you believe you are the victim of click fraud. But, if it meets your ROI, continue to program and address the click fraud issue later.

 

Rescuecom doesn’t want you to talk about your products when…

someone uses a competitor’s trademark as a keyword. Even if you say nothing about the competitor’s product; do not claim to be the competitor nor attempt to deceive the consumer, you should not be part of the results. According to Rescuecom, the unseen use of the keyword creates an assumption in the mind of the consumer that you are, in fact the originator of the product. In other words, you are selling the product they searched on. Although Rescuecom concedes that the practice of associating one product with a competing product off line is a common practice to which consumers are accustom, doing something similar online, they say is beyond the consumer’s comprehension.

One of the presumptions of this lawsuit is that once consumers select a brand or product on which to search, they are only interested in that brand / product and nothing else. I have to say, if Google thought there was any merit to that, they would have that factored into the algorithm…oh, they did.

If  Rescuecom’s (appellate Court pdf) argument has any merit, biding on competitive products and brands will become cost prohibitive. Your quality score will go down, your ranking will go down, your costs will go up and the ROI will erode to an unsustainable level

However, the fact is that is not what happens. In the automotive category for instance, consumers are up and down in the funnel throughout the purchase process. Yahoo! conducted a study last year that showed a distinctly non-linear path to the purchase of a car. I would propose that preventing bidding on competitor keywords is actually a disservice to the consumer (Yahoo!’s policies are ironic). Here, they are seeking to fill the consideration set with comparable products and services. As purchase involvement of any product category increases, so does the competitive set the consumers want to have.

I was further struck by the assertion that consumers would be confused by this form of advertising because they were not accustom to online advertising methods and would therefore be confused into believe the competing brands and products were actually those of the searched brand.

This argument brought to mind the Louis Vuitton case from a few years back. That was when retailer Burlington Coat Factory sold look-a-like bags in an attempt to capitalize on the LV craze. The courts initially said there was no infringement because the two bags could be distinguished when place side by side. This was overturned  in appeals because the bags were in fact not sold side-be-side, thus preventing the consumer from making distinctions between them. Some commentary here.

Rescuecom tried to apply other presidents(pdf) as well. If you review Google’s brief(pdf), I think it is clear that Rescuecom is reaching. The very nature of the online world is access to vast amounts of information. In fact, the search function allows for just what the courts in the LV case originally said was needed to make this okay – side by side comparisons. This just happens to be on screen… with access to vast amounts of information to verify any claims and seek clarity in the event of confusion.

I read Gordon Hotchkiss’ blog today and can’t help but think the folks at Rescuecom are similar to the school administrators he mentioned; very smart, but terribly out of touch with the new realities that the Internet affords us.

Mobile search offers real opportunities

This week Yahoo made two big announcements:
1) They inked a deal with LG Electronics to have Yahoo! Services pre-loaded in LG handsets.
2) Yahoo! is providing mobile display ads in 19 countries using Yahoo! Go 2.0.

The three big rules in real estate are location, location, location. While LG is not the largest, you can’t sneeze at being located on 10s of millions of handsets in 70 countries. The pre-loading brings many users who may not have the inkling to load the apps themselves into the mobile internet world. At the very least it makes it more convenient than loading a competing application. In and of itself though, this is not enough. Yahoo! needs to have content and applications that users will come back to. I think they have both.

Yahoo! has about as much experience in content development and delivery as any company on the web today. From international to local, they cover it. And, in a world where well targeted advertising is also content, they have the infrastructure to monetize the apps while enhancing the user experience with relevant, paid information.

OneSearch, part of the Yahoo! Go 2.0 mobile applications, has enhance search functions that can be driven off mapping, grouped logically and shared with other Yahoo! Go 2.0 users.

A run down of the features of Yahoo! Go 2.0 include: (From Yahoo! PR)

-- Search reinvented for mobile consumers - oneSearch recognizes the intent of a search term and presents relevant content - not just a list of web links to PC sites - on the results  page. Results are grouped by subject making it easy to read through and drill down to get more details.

-- Be local, no matter where you are - The Local & Maps widget  gives consumers quick access to comprehensive local directory  information for businesses across the US, enhanced with  ratings and reviews from the millions-strong Yahoo! community.  Interactive maps feature the ability to directly search for local businesses, get driving directions and real time traffic  updates.

-- Rich, highly personalizable content from millions of sources - Current headlines and a breaking news ticker in the News, Sports, Entertainment, Weather and Finance widgets keep  consumers connected to the information that matters to them.  Consumers can customize to receive content from the millions of sources on the Web that publish in RSS.

-- Photo sharing keeps consumers connected to their community -  The Flickr(TM) widget integrates one of the Web's most  innovative and prolific photo-sharing communities, making it  easy for consumers to upload(1) and manage images from their  camera phone. Consumers can also easily share photos, view their friends' pictures and browse or search the millions of images.

-- Streamlined e-mail keeps consumers in sync - The E-mail widget allows consumers to quickly respond to, delete or compose new  messages or view attachments with a single click - all  automatically synchronized with their Yahoo! Mail account in real time.

What is important here is relevance. This point was made well in the Jim Leggiere interview with Adam Soroca of JumpTap, where Adam said “In practice you've got targetable information that's more locally based and actionable and timely”

I know branding is going to play out here. In fact, the initial launch of Yahoo! Go 2.0 will have display ad support in 19 countries. This is Yahoo!’s initial focus (Yahoo! PR 2/12/2007):

“..This worldwide launch demonstrates Yahoo!'s continued focus on extending our leadership in display advertising and intension to be number one in mobile monetization.

Major global advertisers including Hilton's Embassy Suites, Infiniti, Intel, Nissan, Pepsi, Procter & Gamble Asia Pacific and Singapore Airlines will launch the inaugural mobile advertising campaigns on this new platform beginning today. “

However, the real value for mobile search will be in the localized, relevant nature of the content (again, Adam’s view). If Yahoo!, Google, MSN (and others) can create the right mobile environment, while delivering targeted search content, there is a tremendous value to businesses. (Danny Sullivan did a piece on the mapping applications on 2/12/07 – demonstrating that there is room for improvement.)

The first wave of beneficiaries here should be the restaurant, auto repair, hospitality and like industries. These are logical mobile searches (I am not convinced that in depth research on “not-needed-now” goods and services are going to be the bread winners here.) What is intriguing about the Yahoo application is the ability share the information easily. Want to let everyone know where the restaurant is – no problem; search it, map it, send it. When you need to have your spouse or friend pick you up at the repair garage to which you were towed, map it and send it. The point is, these applications, and other mobile applications facilitate a mobile lifestyle and incorporate a behavior (search) that is already established. I, like many others, see this as one of the great next waves in search marketing. While 10% if the mobile hand set users use mobile web, that number will grow as will the types of searches made. We need to get our learnings in now.

 

Personalized Search. For whom?

Google announced the launched their new search algorithm that takes into account a user's history of clicks, bookmarks and other behavioral factors that help narrow the links to what Google believes is most closely related (if they are signed in, which is my issue). On the Surface, I thought this was great. I am very different in my intent for a search on cars than most 40 year old men. I have had clients in this area for more than 15 years, and most of my interest is in the marketing of automobiles and the industry vs. the mid-life-crises-induced-hunt for a cherry red Vette (don't get me wrong, I've done that too).  So, if the information that relates to new ad campaigns, change in ownership of aftermarket repair shops, or dealerships is prioritized over new model promotion (still of interest but much easier to find), that would make my life much easier.  

 

However, I use our home computer for work. My kids and my wife use it even more. Our oldest son would likely be looking for pictures of cars he has seen on his psp racing  game.  My youngest would be looking for pictures and clips from the movie "Cars."

 

 If am doing research for work, or shopping for a new car, neither of these histories, behaviors or bookmarks has anything to do with my intent.

 

Okay, so I should log into my account, then when my wife wants to surf, she should log into her account. Then, my kids should do the same. Of course, I can just buy every one a computer. The truth is, until this process of identifying users becomes nearly automatic, I don't see the Google method as adding a great deal of value.

 

So, I am conducting  my own unscientific survey to see if my situation is typical, or if I am the  "focus group of one" I often rail against. I’ll set this up and post it.

 

Google: Spock in a James T Kirk world

I thought the Google entry into the general media world was approached by the Googlers I met as a bit too simple a process. Almost a foregone conclusion that the agency world and advertisers would jump on the Google express.

I was interested to read the mediapost piece  on the dMark founders' separation. It appears that there is a disconnect at Google on the other side of the equation as well. When I talked with some folks at Google about the agency relationship and that they would not want to insert technology between them and their clients, or rely too much on Google's technology, they did not agree. Well, it would appear that there was an internal discussion going on about the same thing:

'...the two companies apparently differed over the need for a "human touch" in the sales process.'

You can not automate good relationships. Further, sometimes, against all logic, people do not move based only on the numbers. The human touch, direct and personal, will trump automated email, IMs and algorithms every day (this was not as evident in search because humans could not do what algorithms could). I see this being adopted at Google more now than just a year ago. The difficulty they will have is that the next steps in the growth - off line - will be much more challenging, and face a broader range of competition than their search engine efforts. The human touch will be key to making it work for them.