Maximizing Your PPC Effectiveness with Brad Geddes

In this Episode

• [01:03] – What are some of the biggest mistakes Brad sees people doing with AdWords or with PPC in general?
• [03:29] – Brad talks more about audience targeting.
• [05:22] – For people who aren’t familiar with the term “lookalike,” Brad defines what it means through a clear example.
• [07:11] – Google and Facebook are the two largest engines that offer lookalike audiences.
• [08:27] – Brad defines “retargeting,” then talks about the concept in more detail.
• [10:23] – If you want to have multiple kinds of ads (such as display advertising, search ads, and retargeting), what should the mix look like?
• [13:34] – Brad gives advice on how to proceed if you want to target individuals but you don’t have their email addresses.
• [14:21] – Is there a time limit for how long a remarketing audience stays, and does it vary by platform?
• [16:32] – Brad talks about what scale means to him. He clarifies that to him, it’s a minimum of 5 to 10 times what you have now. He then goes on to talk about ways of scaling properly.
• [20:42] – Brad takes a moment to define some terms that may be unfamiliar to listeners, such as responsive ads, lightbox ads, and in-market audiences.
• [24:16] – What are Brad’s recommendations on how to design display ads? He admits he’s a terrible designer and doesn’t even have Photoshop, but then explains that Google has a way of making HTML5 ads.
• [27:00] – Brad explains what dayparting is, and talks us through how using it can be an effective strategy.
• [29:25] – We learn more about negative keywords and broad match.
• [33:11] – Brad discusses various softwares that are relevant to the techniques he’s just been describing.
• [35:08] – We learn about different kinds of negative keywords, and how to use each type. He then talks about the number of negative keywords across these types.
• [38:50] – Stephan and Brad talk about ad groups in AdWords.
• [40:11] – How many ad groups would Brad have with a typical client?
• [41:50] – Brad addresses the topic of when it is (and isn’t) worth it to buy your brand name as one of your keywords.
• [44:44] – Brad clarifies the scope of what he does. He then addresses Stephan’s question about doing audits for new clients (in terms of time and cost).
• [46:28] – What would someone expect to spend if they hire an agency? After answering, Brad offers recommendations on how often to check on the performance of the agency you’re working with.
• [49:13] – We hear Brad’s thoughts on dynamic insertion ads. He then shares his favorite ad customizers.
• [50:47] – Brad speaks to the importance of congruence between the ad and the landing page.
• [51:52] – Brad runs through the ad extensions available for AdWords.
• [55:07] – Other than AdAlysis, what products does Brad like?
• [57:02] – Brad lists some of his favorite online tools, explaining what he likes about each one.
• [58:56] – For ad testing, what are Brad’s favorite platforms and procedures?
• [60:58] – Brad shares his thoughts on YouTube ads, and how they compare to ads in other places.
• [63:11] – What are the next steps if someone wants to connect with Brad?

—

Transcript

Hello and welcome to Marketing Speak, I’m your host Stephan Spencer. Today we have Brad Geddes with us. Brad has been involved in PPC since 1998, he’s a really an early pioneer of pay-per-click. He’s co founder of AdAlysis, an ad testing and recommendation platform and Brad is also the author Advanced Google Adwords and an official AdWords seminar workshop leader. Brad, it’s great having you on the show.

Thanks for having me, great to be here.

Let’s dig into Adwords and PPCs. First of all, what are some of your biggest mistakes and mess ups that you see out there folks are doing with their AdWords or just with PPC in general?

It’s amazing that this is still a common mistake. Number one, and I’m sure you see this in organic too, is not collecting your data properly, whether it’s not having analytics or conversion tracking installed. I think it’s only 46% of companies actually track transactions, that’s a clear number one. Number two is just not getting the fundamentals right, it’s searching for the newest, coolest, latest thing to implement and not just getting to the basics of this is a keyword, this is an ad. With these two, my landing page work really well together, I have a good fundamental ad group and I need to repeat that before I move onto all the cool audiences and this and that. When you get into scale, if you can’t get the basics right, nothing else matters. For those who are a bit more advanced and doing this for a while, it’s really not understanding audience targeting. Years ago, people would always look and say, “Well, what’s an average conversion rate? Give me a benchmark to start from.” 2% to 4% was always a pretty common, legitimate number to use. That number, it’s not that it’s old, it’s still true but it doesn’t reflect the current way people interact with websites. The average person will visit the site six times in a transaction process. It could be six visits to one different site each, it could be six visit to your site and so forth. Not understanding the fact that paid organic put together, organic in email and paid all play together. The fact that it’s not just get them to your site, it’s get them to your site, watch their behavior and then determine how to market back to them until they decided and say, “Hey, I’m going elsewhere,” they’re not converted into anything or, “Alright, let me become a customer.” That’s probably more advanced audience based targeting, whatever you want to call it really is what the advanced users are getting but not all of them are really going to the extent they can.

Got it. Let’s talk a bit more about audience targeting. How does that all work? What are some of the most important things to know about it?

Audience targeting, essentially, at its heart is just saying a user meets condition X, I wanna market to them differently. Condition X can be based on the fact that they’re in your CRM, they might have been on your website before. They went to your trade booth and you collected business cards then uploaded to your CRM System and these are trade show attendees. An obvious example for an ecommerce system is someone who abandoned your shopping cart is different than someone who just went to your homepage for three seconds and left your site. An audience at its heart is just a group of like behaviors. That’s why it’s conceptually right, it’s very simple. But when you start saying, okay, what groups of behaviors exist that we wanna market to and then how do we market to them differently based on that behavior? If someone filled out a lead gen form and you’re sending to a sales rep and those sales reps try to call them back on the phone, that’s where reinforcement about sales reps in your marketing. Where if someone has abandoned your shopping cart, hey show them products in the shopping cart. If someone is a high prospect form a trade show you went to, that’s a completely different marketing message. It’s really what are the behaviors that your customers go through, what are their like behaviors that you can market to and then you can do it based on website behavior, you can do it based on lookalikes from website behavior, you can do it based on email addresses, there’s a lot different ways to do it but at its core, it’s identifying behaviors that you wanna treat differently from the norm.

Let’s define look alikes for folks who are not familiar with that terminology.

Sure. A lookalike audience is an audience with similar characteristic to another audience you created. For instance, let’s say that you’re an ecommerce site and you sell high end dog supplies so you make a list of who abandoned a shopping cart. This is probably people who are pet owners which are often people in families and who are pet owners who obviously are on the higher income range if they’re looking at high end dog supplies. These are the characteristics that if somebody has, they’re like this audience that you specifically created. Now, let’s find people who have the same characteristics and build you a look alike. It’s not first party information, it’s third party information. Conversion rates and click through rates aren’t as high as first party but it’s much higher than not having any kind of information about a group to begin with.

The look alike is based on an algorithm that the platform creates, like Facebook is in charge of creating that look alike audience and not you, you feed the data in and then it goes into this black box and then another audience comes out.

That’s fair, yup. That is correct. You don’t have the data because if you have the data, that’s first party information. This is taking your first party and then looking at a third party data and saying what looks similar, characteristics, and then creating that look alike, the engines do that for you based upon that.

What are the engines that offer lookalike audiences?

Google and Facebook are probably two biggest. When you get into display, a lot of the display providers offer lookalike targeting as well. Any larger online advertising systems at some point in time has lookalikes. Bing does not yet, they don’t have much of a display network though. Facebook, Google are the two big ones from us performance based marketers but when you get into display buying, most of them have lookalike type capabilities as well.

Let’s now jump to just the general concept of retargeting. There are lots of ways that you can retarget, you can retarget with Google, you can retarget specifically with YouTube, you can retarget with Facebook. First of all, let’s define the retargeting, remarketing terminology for those who aren’t familiar with it. What are some of the use cases, you alluded to that a bit already with saying take your email list or a list of people who abandoned at the checkout process and that’s a custom audience that you can target with specific ads.  

Retargeting or remarketing, it’s remarketing in AdWords, it’s retargeting with any other display provider. It is a subset of audience targeting where you define if a user does X on our website, put them in an audience list and then serve ads to that user based upon this behavior on our website. The most common, of course, is online checkout abandonment. That’s where you go to Best Buy site or go somewhere, add stuff to your cart and you leave and see you ads and say, “Hey, there’s stuff in my cart.” You wanna come back and buy it. That’s shopping cart abandonment. You can do remarketing based on the products a person viewed. It doesn’t have to be ecommerce. Products a person viewed is a ubiquitous statement. If I go to Expedia and I look at plane flights to London and I look at three hotels in London, those are products. Therefore, they could dynamically build an ad that shows, “Hey, did you know prices from DC to London right now have dropped to this and here’s the prices to those three hotels you looked at.” It doesn’t have to be straight ecommerce, it could be anything that’s product based, it could be courses at a university. Remarketing is really serving ads to a user based on specific on website behaviors and products viewed.

Now, let’s say that you wanted to have multiple kinds of ads that you put out there, you wanna do some display advertising, you wanna do some search ads, you wanna do some retargeting and so forth, what’s the next look like? Should everybody be doing display? Is that where you’re gonna get the best value or why is it more the search ads, is it Facebook advertising?

There’s search retargeting and there’s display retargeting. With search targeting, the user has to search again for something you offer and then you could say, “Oh look, you’re an audience, you’re on our site, let’s do this.” It could just be increase a bid, so forth. That inventory is much smaller because the user has to initiate that search that still triggers an ad of yours. Where Facebook or display based advertising, AdWords, is a much larger impression share because the user just has to have done the behavior on your site because you’re not targeting the user, not what the user is doing on this other site. If you think about this a search engine perspective, let’s say that I wanna buy an ad on New York Times, it’s a tier one property, it’s got a lot of traffic, that’s gonna be expensive to buy regardless of the user I’m reaching. Let’s take ‘johnnysblog.com’, Johnny’s blog does not have a premier advertising price to it, it has a really low reserve price to it. From a search engine perspective, suddenly they can say, “Oh, hey look. This primo user of yours is on Johnny’s blog.” You actually wanna pay a lot even though it’s on randomsite.com because that’s for you. From an engine’s perspective, they actually make more money and you as an advertiser now make better usage of second and third tier inventory than normal as well. There’s a lot more impressions on display possible, but with Facebook, because it’s a limited adslot to begin with, there’s a lot of people targeting every condition on Facebook, often it’s a higher CPC retargeting in Facebook than when you get into second and third tier websites just randomly out there or apps or what not because they have much less competition for ads and because the advertiser doesn’t care about the site’s quality, they care about the user quality. Suddenly, you actually can afford and want to be on some of these sites because you’re reaching the user who then is more likely to come back and buy from you and engage in your services.

You’re saying that Facebook cost per click can be higher than on AdWords if you’re doing it in a way that is smart, going after the second, third tier in what you’re just describing.

No, AdWord search is more expensive than Facebook. But when you get into the straight remarketing audience targeting, Google is usually cheaper with their huge display network than Facebook is, just to layer them out there.

Got it, that’s a great distinction. What if you don’t have email addresses of individuals that you wanna target? You wanna target individuals, you don’t have their email addresses, what options do you have?

Start with your website information. If you’ve got Analytics installed, you can start looking at here’s our website behavior and then there’s our website behavior, these are the ones we wanna reinforce, wanna market back to people. You need as little as 50 people in a six month period of time to do a behavior to be able to remarket back to them. If you can’t get that then you need to be at the basics of just building a web presence anyway. That’s how granule you can start to get as well.

Is there a time limit for how long a retargeting audience stays and does it vary by platform like is it 6 months, is it 60 days?

It varies by platform. Generally with AdWords, you can do 1 day to 540 days which is longer than any cookie expires anyway. You gotta think about why you’re doing this. We just passed December. There are websites, think about Harry and David, they make half their revenue basically for the year, I forget if it’s 40% or 60%, but it’s a significant amount, between Black Friday and Christmas. They’ve got this group that buys a lot in that holiday, they’re not gonna buy for a year, they’re gonna, in their case, collect email addresses and say, “This is our previous year buyer group.” Because you’re using email addresses in that particular taste, there’s no cookie to expire, you’re re-reaching the same users who maybe for a year didn’t care about you but it’s now the time of the year they do. Because cookies as you know, the odds of a cookie lasting for two years is pretty slim but you can theoretically set them for that long.

Let’s say that I wanna target specific individuals that have identified through Facebook user IDs or their URLs on Facebook. I don’t have their email addresses, can I target them or is that just not an option?

I am not positive to answer, I am not a Facebook guru by any means, I do mostly upload remarketing off of Facebook and I don’t go fully into their ad platform because it’s so convoluted as it scales as well as other platforms do. I actually don’t know the answer to that question.

Okay, let’s talk about scale then. Let’s say that you wanted to scale an existing campaign and spend, what does scale means for you? Is it going from let’s say $5,000 a month in AdWords spend to $50,000. Is that scale or is it more like 10 to 50 times increase in budget or is it more increase in some sort of other outcome metric?

Scale is basically saying you’ve got something working and you want it 5 to 10 X now, minimum. The larger budgets we work with are between 10 and 12 million per month per platform. There are some we deal with that they spend 10 to 12 million on just AdWords and they can’t spend it on Bing, of course. That’s only a million in Bing and another 3 million on Facebook. There’s not many companies who can get that kind of scale, that’s a crazy amount of scale. Going from $5,000 a month to $25,000 or $50,000 a month, that’s scaling what you currently have. Scale is really just saying we wanna do this, we want to 10x what we are now.

How do you do it? How do you get that scale? As you crank up the spend, don’t the results start degrading if you buy a lower level or lower quality traffic?

Well, not necessarily. This is ways of scaling properly. Number one is to really take a look at keywords first, you’re talking like in AdWord or Bing, it’s keyword first. Can we get more out of our current keyword set? We’ll see all the time a nationwide dentist company. They’ve got the word dentist, but they don’t have dentists in Florida, dentist in DC, and that’s a geographic based scale that by moving to every individual geo as a keyword and then also replicating that in the ad copy, you theoretically may not grow impressions much but you might grow your clicks 500% because your clickthrough rates are gonna be so much better because you’re addressing a local based level. If your landing pages correspond to that, you’re gonna see that nice increase in conversions. Scale doesn’t necessarily just mean adding more, it could be increasing efficiency of current impressions. Step two, do we really have all the keywords necessary and keywords possible? This is again, research, brainstorming, thought process and so forth. It’s usually let’s move into audience targeting. Your first step into display can be scary because if you don’t do it well, you can lose a lot of money on display and not getting many good result. It’s usually best to start your foire with audience targeting because you know something about the group. Once you really understand how to make image ads and responsive ads, Lightbox ads that are speaking to a group you already know about, once you understand this is how our consumer base responds to ads, this is what works, now we can move into similar audiences or in market audiences. We can see if we don’t have first party data, how do these scripts interact and let’s learn from that. Now that we understand, okay this is non-first party but good targeting and that’s how they respond and react, now we can move on. In each level, your conversion rate, your CPAs may decrease a little but you’re potentially adding a huge amount of impressions each one of those steps. The blended CPA should still be within a reasonably target. There’s no bad keyword, there’s no bad targeting, there’s a bad bid for it. If you can only afford a penny for it, bet a penny. You may not get huge scale but it’s at least part of what you’re getting, you can increase efficiency from there.

Interesting. Let’s define some of these terms for those folks who aren’t familiar with them, like responsive ads, Lightbox ads, in market audiences. Let’s define all those.

In market audiences exist only in a handful countries, it’s a Google prebuilt audience type that looks at people currently researching products and services. If you go and search for four different car types, you’re probably in market to buy a new car, you’d be an in market car buyer. Where if you just search for local dentist and cavities and how much does it cost to get my teeth whitened, then you’re probably in market for a new dentist in your local area. In market are predefined audience list of users who are researching products or services online.

What geo is in market audiences available?

United States, UK, Germany, I don’t know if it’s been launched to others or not yet. I know this is something that Google’s working hard on adding to more and more countries but I don’t know the full breadth that the in market expands to in the moment.

Responsive ads, Lightbox ads.

Lightbox ads, they’re a small form that you click on and expands on a page. You might be on someone’s site and you click on the image and image blows up to much larger than normal, that’s a lightbox image, Lightbox ad is something similar. Responsive ads are ads that can take many different sizes and shapes with the prebuilt information. For instance, when you look at ad slots across the web, ad slots vary in size from a 720 pixel by 90 pixel to a 600 by 160 skyscraper to a text ad size. What a responsive does is you upload a logo and image, short headline copy, long headline copy, so forth and then based on the ad format size, the ad is built at the time of being rendered to that specific ad slot size.

Awesome. Do those perform really well?

They perform way better than text ads because there’s a visual component, that’s not surprising. If you know an ad slot size and you make ads for a specific ad slot sizes, they generally perform better because this is a 250 by 250 ad slot and I can get my call to action just right in how it’s all laid out. It’s a lot more work to make image ads for every possible size but if you put in the work, generally they’re responsive. Advanced responsive, you can make a couple ads and pass big messages and they get rendered in every ad slot size regardless if it’s for a small mobile screen or a massive wide screen monitor. It’s time versus energy type of situation there, but they definitely do better than text.

How do you recommend folks design display ads. Is that something you go to a marketplace and get people bidding on creating this ad for you, like 99Designs situation or do you go to an agency that designs these display ads or do it yourself?

Personally, I’m a terrible designer, I’m the only person who’s been in online marketing for 19 years who doesn’t even have Photoshop on their computer, I’m just a terrible designer. I like to tinker with the text, I’m a really good writer. Google has in their interface a way of making HTML5 ads. They’re like Flash, they can have animation and so forth in them. Scan your website and play with it. I like those first because I can look at colors, I can look at what address they’re showing on publisher’s site because some sites a contrasting ad does better, some sites a blended color does better. I can easily make a few different ad types and colors and messages and seal works really well. Once I know that, then I use 99 designs all the time or Upwork or something and say, “Alright, this is what my ad looks like, this message works well. Now, make it actually look like it’s professional.” That’s usually how I approach it. If I need true inspiration, it’s amazing to say, “This is my idea, this is my message, give me some inspiration.” You’ll get lots of good ideas to play from. I’m usually a fan of that. If your company has a brand department, you probably can’t do that, you probably have to go to an agency or creative service to follow brand guidelines and legal approvals, that’s gonna depend on an internal process too.

Have you heard of Banner Cloud that’s like a marketplace and a toolset for designing banner ads, the marketplace includes people who actually design them for you if you want.

I don’t know that one. There’s a lot of good ones out there. Have you tried that one?

I’ve got an old demo of it. I haven’t used it because I don’t really design ads myself. Jon Shugart is one of the past guests on Marketing Speak and he’s the guy behind Banner Cloud. Listeners, check out Jon Shugart’s episode by the way because it’s a great one. It’s one of the earlier episodes on Marketing Speak, he talks about webinar best practices and all sorts of good stuff. Let’s go into some other areas of pay-per-click and some of the best practices and features and functionalities like day parting and negative keywords, a whole bunch of stuff with dynamic insertion, let’s start with day parting, what is it and does it work? Is it helpful?

Day parting is essentially when your ad show or don’t show and you can do day parting with bid modifiers to say increase or decrease my bid at this time. This time could be [8:00] to [5:00] on a Friday, bid down, but on Saturdays bid up, or Saturdays at noon bid up. It’s great because when you think of users, a search on a Monday morning is more likely related to businesses and getting work done for your work week, a search on Friday night is more likely to be for entertainment purposes and finally getting out of the office. The search on Saturdays is more likely to be something like vacations or stuff with the kids and whatnot. When you think of marketing, we’re just reaching people and that’s what’s so easy with paid search is to forget that a click is a person and an impression is an eyeball of a person. Our daily routines dramatically affect how and why we search and what’s useful to us right now. Understanding the fact that in B2B, Sunday nights is actually a huge search night for one who left early Friday and didn’t get their work done, you’re not getting any calls. Monday morning is a great time to get those calls. If you’re selling cruise vacation packages, Monday morning is an awful time to show those. Lazy Saturday afternoons is an amazing time to show those ads. Being able to look at your industry and products and services and say when do people actually need us? Then we should bid up. We’re probably not their choice one, we should bid down, you should do that based on actual conversion, that’s not just guessing at it, use your data. It’s absolutely a good idea to change bids by time of day, day of the week when you have the stats to prove it out of how your conversion rates and CPAs and so forth change.

When you think of marketing, we’re just reaching people and that’s what’s so easy with paid search is to forget that a click is a person and an impression is an eyeball of a person. Click To Tweet

What about negative keywords? I guess we need to define broad match and all that for folks too.

Match types are essentially how closely related a search query is to your keyword in order for an ad to show. Forum basic match types, exact match means the query and your keyword are almost exactly the same. Phrase match means the search query is contained within your keyword or your keyword is actually contained within the search query but there can be words before and after. Broad match means the query and your keyword are somewhat related, I stress the word somewhat because a query and your broad match keyword might actually use all different words and your ad can still show and you’ve got ad modifiers which is like broad match except you’re saying the plus symbol must be really closely related to what’s in your keyword, the rest can be treated however. Negative keywords are for the opposite, they say if query contains this word, don’t show my ad. They’re really, really important because your keywords in most cases will match to some search queries that are just things you don’t do. If you don’t use negatives, you’re gonna spend money on queries that are unuseful to your business, you must do it.

An example would be if you buy shoes on broad match, you don’t wanna buy snowshoes as a keyword or brake shoes or horse shoes or used shoes or any of that. You wanna include all those as negatives.

Yup. From an advanced standpoint, we use some engram analysis a lot for negative keyword research. Are you familiar with Engram’s?

No, let’s talk about it.

Let’s use a really simple example here. Let’s say I’m a plumber and someone searches for licensed plumber, someone else searches for licensed plumber in Seattle, someone else searches for plumbers in New York and someone else searches for licensed plumbing services, four queries. Normally, you would look at these four queries independently and say is this good or bad for me? What an Engram says is you don’t have four queries, you have one, two and three word combinations used across queries. For instance, we have four uses of plumber, we have two uses of in, we have two uses of licensed, we have one use of the word New York and one use of the word Seattle. I might look at my Engram then say did you know that when the word in is contained in the query, we don’t do very well. The word in used signifies geography. Maybe we need to think about either making it a negative keyword because it doesn’t work for us or we need to get more detailed on geographies. Did you know when the word license is in the query, we do really well because we’re a licensed plumber. We should split up our queries between plumbers and licensed plumbers because we actually are licensed and bid up when the word license is contained because that matches our business closer. Engrams are a way of finding patterns across queries at once that give you some really good insights, whether it’s new products to add or whether it’s good negative keywords to add, as opposed to looking at an entire search string at once.

Is there a software that helps you to do this?

AdAlysis, our software actually does this automatically, there are scripts you can write which take a bit more work that you can install in AdWords to do an engram analysis for you as well but it’s really useful to find negatives. Just find new insights or I’ve even seen people build new products because they realize that in aggregate, there’s actually a lot of volume for this search term but we never noticed it because you’re normally looking at if you see 10 impressions, you say who cares, when you say 100,000 impressions, all of a sudden it raises your eyebrows, even though that 100,000 maybe across 1,000 queries, you never saw the pattern.

AdAlysis, that’s your platform, does it also find negative keywords?

You can use that as negatives because theoretically, we’re not gonna say a keyword is really negative because that’s dependent upon your cost per acquisition and target conversion rate. One company might look and say, oh wow, $1,000 CPA, we’re doing great. Another company is gonna look and say, oh, $5 CPA, we’re too high this week. We’ll show the engram analysis and the conversion rate, CPA cost, impressions, so forth, by every one, two and three more theoretically to engram, bigram and trigrams. By each of those combinations, a user can just filter and say show me engrams with zero conversions by cost, I’m gonna take the top three and make it a negative if they’re above my target CPAs.

With AdWords, you can have up to 10,000 negative keywords in an ad group. Is there a typical number that you aim for? Do you try and max out the full 10,000?

Why do you know that number?

I’m a geek.

When you think of negative keywords, just three places to add negatives inside an account. One is at the ad group level, that only affects keywords in the ad group. One is campaign level, affects all the keywords in that campaign and those campaign negative list which let you build a list and then apply it to campaigns so your all campaigns can share specific negative list. If I don’t ever wanna show for a keyword again, I make it a campaign negative list. If I wanna show this query from one campaign but not another campaign, then it’s a campaign negative. Ad group negatives are mostly used when you want one ad group to show for something but not another one. Let’s say you’re hosting providers, you got an ad group called hosting, an ad group called VPS hosting, an ad group called cheap hosting and then cheap VPS hosting. The problem here is that if someone searches for cheap VPS hosting, it could be from a hosting ad group or VPS hosting ad group or a cheap hosting ad group. We lost control over which one because we’ve got that keyword, assuming it’s not an exact match, matches all the terms, all our keywords match that one term. We will take our cheap posting ad group and we put minus VPS in the ad group level so that can’t show from that ad group, the other one can. If we don’t do cloud hosting then we take the word cloud and we make it a campaign negative so that none other ad groups can show. I don’t use many ad group negatives, that’s mostly for shaping account when multiple ad groups can show for it, I more likely use campaign or a campaign negative list.

Which has got the longest list of negative keywords for you, typically?

That depends on the account type. There’s one account I’ve worked on for 12 years now. It maybe has in totality 1,000 negative keywords and there’s others where we’ve got a lot of weird shaping going on that it’s 2.5 million negative keywords, 3 million negative keywords. I don’t think there’s a good number, I don’t add negatives because it doesn’t look good. If you start making those guesses, it’s amazing how often words you don’t think should convert actually do. I let the data speak for itself, I remember we’re working on an account that’s software downloads for a very legitimate large company. There was a term that was three licensed, it’s something like acrobat. Acrobat hacks and acrobat thorns, they’re like we should add thorns and hacks and free as negative keywords, I’m like, you’re guessing on here, we looked at it and it was converting. What it is is that you’re not addressing this correctly. They made an app that said, cost of software’s name is $79, cost of cleaning up malware downloaded from Torrent sites $3,700, what would you prefer? It’s not the word was negative, it’s that they didn’t address the need correctly in the ads.

That’s great. With ad groups, you wanna really think strategically so that you’ve got the right size buckets and you’re not having words end up potentially firing off of multiple ad groups. You gave that example earlier of licensed plumbers and those licensed keywords would be in its own ad group.

Correct. Assuming that they make a big difference, there comes a point that you can split ad groups for no performance gain, you’re just making more time which is why I like to let the stat show me which way to go because it’s too easy to end up with one keyword, one ad, one landing page and that gets to be more management than it’s usually worth. Start logically granular, if you got a Seattle or a New York keyword, do not put them on the same ad group because that ad can’t reflect than. When you continue breaking things more and more like plumber versus licensed plumber, you wanna make sure that stats bear out that it’s actually worth while to break that up.

Be pragmatic about it, don’t create a whole bunch of extra work for no extra gain.

Yeah, it’s easy to do. It’s busy work, no one likes busy work.

In a typical account of a client, how many ad groups would you have?

Again, that so depends on what you sell.

Let’s say it’s a one product company like Bowflex, then we’re talking about even in that case hundreds of ad groups?

We’ll take Bowflex for a second, everyone’s seen their TV commercials. It depends. From my brand campaign, I refine one ad group Bowflex, pretty simple. What do we find about our conversion rates? When someone searches for better abs or one room workout or foldable exercise, we see those are different then we suddenly need to have an ad group that talks about, hey, this is how it folds up versus this is how you target a specific body part in your workout. Those would each be independent ad groups and that could grow to a few hundred ad groups very, very quickly. If you look at it and say everyone knows Bowflex and the keyword of high performance abs versus single room workout devices are the same with the same ad, then don’t waste your time. Odds are it’s not gonna be in that case, but that’s one extreme example. I get to work on accounts where a million ad groups is not a known number and others where 50 is all they need.

Let’s say that you’re buying your brand name as one of your keywords and that can be an unnecessary expense. How do you know when that’s a necessary expense or unnecessary expense?

Number one, if there’s another advertiser on your name, it’s usually worth it, but without doing an experiment. You can run the paid and organic report. If you got a Google search console account and you connected to AdWords, you then have access to a report that shows you for this keyword, this is your clickthrough rate and your stats when only your ad is on the page only your organic is on the page versus both were on the page. You can look at the report, it’s available for you if you connect those two things together. To make that decision, when we have our brands twice in the page, our CTR is even higher, paid and organic combined, versus not. Or hey, our organic listing versus our adipose organic listing, same CTRs, let’s stop buying and see what happens. I’m a stats person so take your webmaster tools or search ideas, then they rename it so often. Take your search console, connect to the AdWords and you can actually see a report that shows you the differences right inside of AdWords.

Years ago I read an article for search engine land about organic and pay-per-click, is it cannibalistic or synergistic? I included a case example of a client who inadvertently turned off their AdWords, somebody screwed up and the organic lift in traffic made up for all that paid search that they had stopped spending without meaning to.

Sometimes it does, it rarely happens but it sometimes happens. The caution that we also have there is that if you’re gonna buy it, penny cheap, cheap buy. Don’t send the traffic to your home page, your organic traffic is going to your homepage, you just know how to click on your logo, go to your homepage. Send it to a page you wish more people knew about, whether it’s a newsletter sign up page, whether it’s a what’s new in our company, product’s page. Get additional value out of that couple pennies since you did pay for it as opposed to sending it to your homepage.

That’s a great point. That’s one action item for our listeners to take for sure. Let’s say that you come into an existing account, they’ve got AdWords going for years and you’re doing first an audit looking for what’s missing, the misconfigurations, the screwups and then you’ll take over the account after you’ve identified all the misses. What kind of length of time does it takes to do that kind of an audit and what’s the cost, do you just do that as a stand alone offering without taking on the account?

We’re consultants, we do not do ongoing management outside of some long term legacy client which is why we help people actually onboard agencies or go from agencies to in house because we’re totally independent, we don’t want it. The time and cost really depends on account size. If someone’s got an account that it’s two campaigns, it’s 500 ad groups, it’s gonna be like $1,500 or $2,000. If they’ve got an account that is 12 actual accounts, it is 3,000 campaigns, it is 2 million ad groups, we might do a sampling of that that could be replicated template wise but that could be $20,000 to $50,000, then we may find some common room to work with them in the middle. That is so dependent on account size and sizes of accounts are just everywhere. The largest account I have worked on is 176 million ads, that’s not the keywords even or anything else, just the ads in their accounts, 176 million. That’s a crazy number compared to the average which is probably 2,000.

The scale of that just blows my mind.

That’s why when you talk scale, it’s about the basics and then replicating it infinite times and not losing your consistency.

Let’s talk about agencies for a moment. If someone who’s looking to hire an agency, what would they expect to spend? Is it traditionally a percentage of the media spend or are there multiple models and what’s the typical percentage?

There’s a lot of models out there. In the end, you’re going to an agency because you’re paying them for their expertise, they’re not non-profits which a lot of people think they are, agencies are for profit companies. Percentage spend model is still gonna be translated for them into how many hours it takes or what not. Fees are generally a sliding percentage of spend. For instance if you spend $100,000 a month, you’re gonna pay 10% to 15%, you spend $1 million a month, that’s too much, you’re gonna pay 5%. At $1,000, you may be paying 35% to 50% of spend at something that low. There are fixed fee companies but if it’s a fixed fee, note that they’re again saying, “Alright, you’re paying us $1,000 a month, we want $250 an hour, you get 4 hours a month and that’s it. That may sound good if they’re spending $10,000 a month, that’s a terrible deal if they’re spending $100,000 because you’re not taking the time to work on your account properly. There’s all kinds of models, some agencies only charge a percentage of increase revenue but each side scales properly. They take your account from $1 million to $5 million in revenue, they get an appropriate reward, you get the increase of a percent of revenue. There’s a lot of models out there but at the end if your agency is performing, it’s worth paying them. If they’re not performing, it’s worth firing them too, it goes both ways.

How often would you recommend doing a review of the performance of an agency? Yearly or quarterly or what?

When you start, every three to six months, after that then annual reviews are good. Agencies get myopic at times, just like one person does looking at the same data, you get myopic at times. They have certain accounts they work on, they understand, but then communication gets lax. The agencies that are the good ones to work with are ones that understand you’re hiring a third party not because you question their work but because you’re bringing more ideas to the table they can work with and that second look. We have a couple agencies, we do audits for some of their clients all the time and they don’t care, they know they’re not losing the business. These clients, they want another viewpoint and that’s fine, that’s just paying for yet another third party view to get new ideas and bringing new things to the table.

Let’s do bit of a lighting round here for the next ten minutes. Dynamic conversion ads. Are those great? Are they not so high performing? What’s your take on that?

There’s now ad customizers. Ad customizers can insert the keyword into the ad copy, it helps a little bit, but at a mass level of handwriting, handwriting beats it. You could do countdowns, you could say if they’re on a desktop, show this ad copy, if they’re on mobile, show this ad copy. Customizers are totally worth it because you’re getting scale from that same ad text. But in keyword insertion, just adding the keyword, it usually is not worth that much.

What are your favorite ad customizers?

Mobile versus desktop, huge. Geography, really useful. If you’ve got expiration date, such as sale ends at midnight or this TV show airs Tuesday at [7:00], whatever it happens to be. The countdowns are fantastic if you say, “Hey, seven days left, 40 hours left, 12 minutes left.” That’s a great one to use to put that urgency sense in there.

You wanna probably reiterate that same urgency on the landing page.

Of course, definitely. Make sure they’re synced. It’s the worst that you can ever see, when they’re not in sync and you use server time or something, you forgot your servers on central and you’re doing it East Coast countdown time or something.

Yeah, that can be embarrassing. Any kind of pointers regarding the congruence between the ad copy and the landing page?

If you put an offer in that add, it better be on the landing page. If you use a call to action in your ad, use that same call to action on landing page. That artistic effect is great, user clicked on your ad because of an offer, make sure the offer exist front and center on the website. If they can’t see when they get there, it’s invisible. Make sure that they all fit together.

As far as landing pages are concerned, don’t use your homepage as a landing page for your ads.

At the most part, no. If you’re a local dentist, then that may actually be your best page. If you’re selling a product, then you should use the product base page. Homepages are actually sometimes okay to use but make sure that it still answers the question users search for and what was in the ad. In general, you don’t want to, but there are some exceptions.

Let’s talk about the ad extension that are available for AdWords? Let’s just run through site links and the phone number and all that.

It’s 13 now. You’ve got sitelinks, call outs, structure snippets, location extension, call extension, review extension, affiliate location extension, there’s so many now. Price extension we show on mobile devices, the message extension, you can sms, write from an ad copy now. Everyone should be using pretty much three to five extensions per campaign, minimum. Site links are links that are short independent pages in your website, you should use them. Callouts, everyone should use it, small bullet point true statements. Snippets, you should have two snippets per campaign which can be things like brands you sell, they’re just quick bullet point list facts about something. If you’re a lead gen, you should use a call extension. If you’re a lead gen for an under 24 crowd, you should use the message extensions so they can text you directly from the ad copy. If you have a physical store,  use your location extension. There’s three to five every one, minimum, should have per campaign but they have those three to five outside of call outs, structure snippets and site links. The others might change a little bit depending on your business type and if you have physical locations where you sell on physical stores but you yourself aren’t, whatnot.

You probably wanna combine this with day parting in that so that if your call center is only open certain hours, you’re not buying those ads driving people to make a phone call when you’re not actually there to answer it.

You can day part an extension but not your ad. You could have an ad that brings to your lead gen form during night time but then when your call center is open, then the call extension shows up where someone can call you or go to the lead gen form page. You can day part ad extensions independent from when the ads run.

Great point. Have you tried Facebook messenger ads?

No.

Okay, that’s something to look into, big opportunity there. You talked about SMS messaging and phone calls to call centers, Facebook messenger can really perform well.

I’m on Facebook about every other month. I’m the only person in planet who still, I don’t even have messenger installed on my phone but I only ever look at Facebook on my phone so I actually can even get Facebook messages, there’s actually not a way to do it without going to the website. I’m behind the times on Facebook, I don’t care.

For listeners who do care about reaching folks, especially if you’re further down the funnel, people have already expressed an interest, maybe they didn’t check out from the process and you’re doing some Facebook advertising already, Facebook messenger ads, incredible opportunity.

I’m more likely to be on Twitter.

What’s some of your favorite software other than AdAlysis, your software platform, what other tools do you like?

Google Analytics, Google Data Studio, unquestionably two fantastic products. Great, great product. If we get into bid management and it’s mass amounts of small accounts, [00:58:49] is incredible. If you’re doing bid management for things where you need to do very custom algorithms, then Adobe [00:58:57] can be really useful. Keyword research, I like Google’s Keyword Planner tool with Semrush. If we get into things like competitive intelligence, then SpyFu is good, Semrush is good. At a much higher level, AD Guru is really, really good. If I’m creating image ads or I’m looking for image ad comparisons, you’ve got AdBeat, you’ve got Mote, there’s one other one whose name escapes me at the moment. I’m a fan of tools but my absolute favorite search tools are my calendar, my project management software because when I don’t know what to do, then no tool helps me out. Those are the two tools that actually tell me which tool I need to use today. Everything with me starts with calendar and project management, then it spreads into what is the necessary way to reduce my time load with the appropriate tools.

You’re a real productivity ninja, I’ve really picked up on that attending your presentation at SMX East and you showed some of the ways that you manage your time and attention and organize your life in business and all the accounts that you manage, it’s pretty darn cool what you do. I wanna give you a kudos for that. Any kind of favorite resources online that are not exactly tools like behave.com, any other favorite resources?

One of my favorite is actually Feedly because I can dump all my authorized feeds in one place when Google reader went away, I needed a place to get my information. SearchCap which Barry Schwartz puts together everyday the top stories around the web, I find utterly amazing because there’s something amazing in Behave or SMX wrote something great that is a must read, it’s in SearchCap. If I only got one newsletter, that would be it because it’s in there if you need to know it. I think with Google, is an incredible resource for the high level thinking standpoint, they do an incredible job of really thinking at a high level about marketing strategy and consumer behaviorism. There’s so many good tools out there but mostly I live a lot in Feedly and Nozzle which is my Twitter aggregator, I don’t know follow every tweet, nobody could but it connects to your personal feed and says, “Here’s the important that happened on Twitter that you should actually ready,” not just the general crap that happened out there.

Love it. Alright, this is a great list. For ad testing like A/B test, multivariate tests or landing page testing like testing things like calls to action and hero images, price points, button colors, all that sort of stuff, what are you favorite platforms and procedures?

Platform wise for ad testing, it’s us AdAlysis because you could test an idea or concept across 100,000 ad groups in 10 seconds or you could test individual ads within an ad group. It is automated and there’s so much cool stuff you can do. When we have no competition at that multi ad group level out there, not to brag but we actually do process a billion ads a day or something. When it comes to landing page, Optimizely, I just keep going back to it. We’re doing full funnel testing, LiveBall is amazing, it’s expensive though. For most people, Optimizely is great, I keep meaning to look at Google’s newest content optimization, whatever it’s called but I haven’t gone really into it enough to compare it. You’ve got visual optimizer on Optimizely, they’re both good, they’re both a little bit different, I find them both very useful. HotJar is another one we use quite a bit as well for recording snapshots of how things go and so forth. I really gotten into Hotjar the past year as well as a landing page analysis tool.

How about Unbounce? Have you ever used that one?

Not much because I know WordPress inside and out. For me to launch a whole lot of landing page test with Optimizely and WordPress templates is no work for me. Since Unbounce is pre-created templates, which they’re great. If you’re not a designer or a coder or what not, Unbounce is really useful. I just don’t need it because I can write WordPress hooks in my sleep, I don’t really need to rely on a landing page creation solution with all the templates that exist.

Okay, last question. YouTube ads, do these perform really well better than Search Ads, is the value there in terms of cost per click?

It’s not search, search has an intent that says I want this information. Nobody compares to searches, leave that as it is. When it comes to YouTube ads, YouTube, when you’re doing audiences, remarketing works fairly well. I don’t get quite as good results in YouTube as I used to. Once the new YouTube watch pages came out two or three years ago, because I can’t form my custom banners and things. YouTube, it works decent but it’s my third, search is number one then it’s audience display targeting and then it’s display targeting so I guess Youtube is four and then I’ll move to YouTube and it’s cheap CPCs. If your goal is increasing brand awareness, it’s amazing. If you’re a direct response company and your stuff does not lend well to visual and instruction then it’s worth trying but not at the expense of some display and audience stuff well beforehand.

There is a great episode on my podcast from last year, Tommie Powers just did a whole episode all about YouTube advertising, it’s really good. If you’re interested in getting some tips and tricks on YouTube ads, he’s been able to scale big time, he gets way better performance with YouTube advertising than with pretty much anything, it’s amazing. Listeners, if you’re interested in YouTube advertising, any of the different ad types with YouTube, definitely wanna listen to that episode.

Just make sure, YouTube over half the views are now mobile. If you’re business to business, make sure you’re really trying to do YouTube desktop or you’re consumer based behavior and mobile works well for you then great. YouTube is now 65% mobile based for use.

If somebody wants to work with you, how do they contact you?

AdAlysis and ad testing and recommendation platform, free two week trial, go to adalysis.com, signup, no credit cards, that’s easy to do. If you wanna work with us, then shoot me an email, you can find our contact at AdAlysis or Certified Knowledge or LinkedIn or Twitter. I’m @BGTheory on Twitter. I’m pretty easy to find if you search my name, it’s kinda hard to miss me online.

Brad, thank you so much, you are a wealth of knowledge. It’s pretty clear you are a guru at these stuffs so thank you for taking the time to share all this wisdom. Listeners, I’ll create a checklist of actions to take from this episode. If you want some low hanging fruit to go after, then check out the checklist. That’s all at marketingspeak.com along with of course the transcript from the episode as well. This is Stephan Spencer, your host, signing off, we’ll catch you on the next episode of the Marketing Speak.

Important Links:

Your Checklist of Actions to Take

☑ Track all transactions occurring on my site. This provides valuable (and necessary) customer behavior information.

☑ Focus on detailed audience targeting. For example, if someone abandoned their shopping cart, advertise to them with products from their shopping cart.

☑ Use lookalikes to target potential audiences with more specific, tailored, relevant advertising.

☑ Spend some time taking a close look at the Analytics for my website. It takes only 50 people over a 6-month period to get enough information to successfully market back to your audience.

☑ If I don’t have 50 relevant visitors to my website during this 6-month period, focus on building my web presence instead of targeting to my existing audience.

☑ When I decide to scale, examine whether I’m able to get more out of my current keyword set before I consider shifting to new keywords.

☑ Make a new display ad by using Google’s HTML5 tool to make a draft. Once I’m happy with the concept, hire someone from 99designs, Upwork, or similar to make it look professional.

☑ Use my data to assess when in the day (or week) people are looking for my services. Use dayparting to bid up during those times.

☑ Start using negative keywords to rule out searches that aren’t relevant to the products or services I’m actually providing.

☑ If I buy my brand name as a keyword, don’t send traffic to my home page. Instead, send them to a page I wish more people knew about (such as a newsletter sign up or new products page).

About Brad Geddes

Brad Geddes has been involved in PPC since 1998. He is a co-founder of AdAlysis, an ad testing & recommendation platform. Brad is also the author of Advanced Google AdWords and an official AdWords Seminar workshop leader.