In today’s episode, Amy is talking to herself for a change! Segmentation is something she gets super excited about, so this episode won’t be lacking. She’s talking about what segmentation is, how to get really good at it, and how not to get lost in it.

Highlights & Takeaways:

• Segmentation is what allows you to understand the “why” behind what you’re seeing in your baseline metrics
• Segmentation can seem like an overwhelming concept, but it’s really just about putting your data into groups where there is something measurable or actionable
• There are countless ways to segment your data, which will allow you to monitor behaviors, performance, and other variants
• Segmentation is what allows you to truly understand your accounts and accumulate actionable data, which you can use to optimize, tweak, and grow your paid search accounts
• If you’re still utterly confused after listening to this episode, don’t hesitate to contact Amy for some 1:1 PPC coaching.

Here at Paid Search Magic, we get pretty excited about all things related to getting better RESULTS in Paid Search Marketing.

If you feel the same way, join us here.

You'll get our latest content and podcast episodes that will teach, inspire, and motivate you to achieve new levels of success with Paid Search.

Success! You're in. You'll get our latest stuff soon.

There was an error submitting your subscription. Please try again.

First Name

Email Address

We use this field to detect spam bots. If you fill this in, you will be marked as a spammer.

Get better at Paid Search? Yes please. We won't send you spam. Unsubscribe at any time. Powered by Kit

Show Notes:

Amy: Welcome back to the Paid Search Magic podcast. This is episode 14 and we’re gonna talk about segmentation.

Now our previously scheduled guest is sick, so I’m going to just be talking to you one on one today about segmentation. Which is awesome, because this is something that I really like talking about, and so I get to hog the mic. But we’ll be talking about what it means – what you need to know – about segmentation: how to be really good at it, and really how to not go down a deep, dark hole where everything loses meaning and you surrender to the void.

You can get the transcript for this episode as well as the show notes from PaidSearchMagic.com/14. Let’s go ahead and jump in.

So this episode is about segmentation and I think it’s really appropriate, and maybe even fate, that has us talking about this today. Today on my Facebook news feed, I saw that it’s actually the two year anniversary of when I spoke on this very subject with the Seattle Search Network in the MOZ building. It was a lot of fun for me. I’m not gonna lie, getting the entire presentation together was a little bit stressful, but it was fun to present this to everybody.

The name of the presentation that I gave was “Unlock the Magic of Segmentation Using Just the AdWords Interface,” and it’s kind of funny to me; magic happens to be my favorite word in English. I just love the way it sounds and it has been my favorite word forever, and as I was trying to figure out how to build a brand for paid search, you know, what I really wanted it to be about. I think it was actually putting together this presentation and this deck that had me thinking a lot about magic as it relates to paid search, or what I consider to be magic; just the power of kind of raising the bar and achieving the kinds of results we’re able to get. And so I really started thinking about magic in the context of paid search, and it was just kind of a natural fit.

An extra bit of trivia for you here, the initial idea for this brand was actually going to be PPC Magic and not Paid Search Magic, but PPC Magic, the domain PPCMagic.com, was already taken. So I defaulted to PaidSearchMagic.com rather than paying someone however much money they wanted for PPCMagic.com. And I’m really glad that I did, because in the months and years since then, I think we’ve really seen that PPC is encompassing a lot more than paid search marketing, and a lot of people who do Facebook and only Facebook consider themselves to be PPC marketers. And not to take anything away from that (I’m just still really used to PPC as it relates to AdWords and to search engine marketing), but just because of that potential confusion or ambiguity with those things working together, I’m just happy with Paid Search Magic. Let’s leave it at that.

So moving on, what you’re gonna hear today is loosely based off of that segmentation presentation, but obviously a lot has changed in two years; a lot changes that fast, and especially within AdWords, and so this will be very updated to make sure it’s completely relevant for today.

The reason I love talking about segmentation is because it helps you become deeply knowledgeable about your account.

Now the reason I love talking about segmentation is because it helps you become deeply knowledgeable about your account. Like if all you’re doing is looking at baseline metrics, then that’s really all you’re gonna understand, and you’re not gonna understand the why. But the more you can segment and really get inside your account, the more you’re gonna be able to understand why, and then the more you’re gonna know what to do; you’re gonna figure out how to take meaningful action and really drive account optimization and growth, which is what we want to be doing when we’re managing paid search accounts.

The segmentation is kind of the key to being able to do all of that, just depending on how you have set up and segmented your account, and your ability to use segmentation tools to not just look at everything as an aggregate, but really again, just dive deep, see what’s working, and find ways to optimize or grow that. And see what’s not working and find ways to cut back on that, or to change it so it starts working again. Segmentation helps us to do that.

I want to start off with a story that Malcolm Gladwell tells about a psychophysicist named Howard Moskowitz who was in the business of measuring things, and in the 1970s, he was tasked with finding the perfect level of sweetness to create the perfect Pepsi. And what happened is Howard is running a lot of taste tests, and he’s analyzing the data to try to find just basically a clean, bell-shaped curve to figure out how much sweetness there should be in the Pepsi, and in fact there was no obvious pattern at all.

And what he found was there was basically various sweetness clusters, which led to his “A-ha!” Moment, that we’ve been asking the wrong question; that there’s no perfect Pepsi, but there’s perfect Pepsis. And so basically what that means is some people like their Pepsi really sweet, and some people really don’t, and so to have a Pepsi for each of those segments is really useful. It’s gonna be useful for Pepsi to try to accommodate their audience, and it’s useful for the consumer, that there’s not just like this one size fits all model.

We’re seeing this a lot as consumers, that we have a lot more choices than we used to, so we can find the thing that fits us the best.. You can’t take the smallest and the largest and say everyone’s the medium.

We need different things that speak to us and as marketers, we’re in the business of being able to match a need with a solution that’s provided by the brand or the company that we’re working for, or even our own company. But we need to be able to reach those segments that are most likely to be keyed into those messages, and that we’re speaking directly to those people, and we have options and offers for each of those groups. And so it really, really applies to what we’re talking about here; just the concept of it’s not a matter of “What is the one ad that works the best?” or “What’s the one landing page that works the best?” or “What’s the one bid that works the best?” But it’s about finding each of those different segments and creating a story, or creating a process, that allows us to optimize each segment.

What exactly do you mean by Segmentation?

Now “segment” can be either a verb or a noun; like we can segment something, we can segment data. Or we can have a segment of data. So when we’re looking at it as a verb, like TO segment something, we’re talking about slicing and dicing data into meaningful and actionable groups, and then we’re talking about the noun of a segment, we’re talking about a meaningful, actionable market or data set. So I hope that’s clear. So segmentation is kind of a big word, but this is all we’re talking about, is that we want our segments to be meaningful and actionable.

Now to talk about what “meaningful” means, is meaningful segmentation is gonna help us learn something, either about our audience or about the account, and meaningful segments are going to be grouped by either wants or behaviors. And ideally, there’s usually a reason that we’re grouping those things together; that things are happening the same way, that people have the same wants, or they behave the same way, or a similar way for some very specific reason.

So examples of wants or market segments would be, just using Pepsi as an example, if people are looking for zero calorie Pepsi, that’s going to be a segment, and people looking for full sugar Pepsi is going to be a different segment. So we’re not gonna pitch the same products to both of those groups.

Data Segmentation

Now with data segments, we’re looking at different behavior or different settings, requirements, or targeting. Someone who’s using their mobile phone to access the website is probably going to behave differently than someone who is on a laptop. Looking at what we’re talking about with an actionable segment, basically that allows us to make improvements to the account based on what we’ve learned. And so if we’re seeing, for instance, that the website requires a lot of text entry, we may see that this isn’t really the experience we want to be sending people on cellphones to. So this is kind of an archaic example, because I know we’re all building out mobile-optimized pages now, but just as kind of an old example, we might want to bid down our bids for mobile, because the conversion rate is lower because of the web experience that we’re sending them to.

I just want to share with you an anecdote that I love from The Simpsons, and this is something that has stuck with me just forever, is where Lisa wants to make a hundred copies of a “lost” poster that they’re trying to put up, and so she goes to the copy store and she asks for different types of paper. She says she wants 25 copies on goldenrod, 25 on canary, 25 on saffron, and 25 on paella, and the copy clerk just says, “Okay, one hundred yellow,” and that was it.
So all the segmenting that Lisa Simpson had done to try to get the right look for each of these pages didn’t really matter, because the only thing that was available was yellow. So she had kind of hyper-segmented in a way that just really was not necessarily, and this is something really, really common to search marketers, is we’ll get really specific about segments in ways that either there’s no meaningful action to take, or that actually can harm the account.

We’ll talk about this in the podcast, but it’s important to know that this is one of those Goldilocks frameworks, where you can under do it, and you can also overdo it, and so you really have to apply thought and strategy when you’re segmenting, so that you get the right balance. Because if you go too far either direction, it’s really gonna throw off your data and can throw off your performance.

Performance Segmentation

One of the things I see happening all the time, especially when I’m taking over accounts, is that people like to segment things based on performance that may or may not actually have any meaning in it. A really classic example, that unfortunately somehow is still used all the time, is segmenting by geography. Now there can be a lot of really good reasons to segment by geography, so please don’t take away that I’m saying “Don’t segment by geography,” but what will happen is that people will look … For the United States, they’ll look at performance and volume for each state of the United States, and they’ll say, “Well, you know, what we’re really seeing is the top performers are California, and Texas, and New York, and Florida. So we really need to focus our efforts there, because those are your best performing states.” Now what do those states have in common? You have east coast states, you have west coast state California, and Texas. There’s no real similarities except for this: Those are the largest, most populated states there are in the United States.

(Image source:  Wikipedia)

So if all we’re really segmenting by or looking at is population, there are a lot better ways to segment an account or performance than to look at where there’s the most population. It makes a lot more sense to look at where’s maybe the click-through rate the best, or more commonly, where’s the conversion rate the best; you know, where might the offer be doing the best, as opposed to just where’s the most volume. Because if you’re able to target the entire United States, you can get your volume, and so let’s focus on where the results are the best, not where the volume’s coming from when we’re segmenting by state.

Another place I see people segment quite a bit, or try to, is day parting. Like there’s this really “Because we can” slice and dice; we can say “How did 11 AM do compared to 5 PM?” And there’s a lot of reasons to do that, there’s a lot of reasons to look at time-based segmentation, but too often people will just see something random. They’ll see “Oh, well 5 PM got eight conversions and 11 AM only got two conversions, so 5 PM is better and 11 AM is worse,” where they’re looking either at a very small timeframe, or something that there is just an outlier.

There’s a lot of randomness in our data and we have to be comfortable with randomness, and comfortable with the idea that especially if we’re looking at very small numbers, we’re gonna see just a lot of natural variance, and that’s just how things work.

And so what I don’t want you to do is to go in and say, “Well what’s big and what’s small?” and “Let’s optimize and give more budget to the bigger and less to the small.” Because unless we actually are confident there’s meaning behind that, then it’s really not serving any sort of purpose and it can really start throwing off the data if our bids are just really out of whack and not in harmony with how the account actually performs. So that’s something to keep in mind, again especially if you’re just kind of desperate to get those conversions, or you’re just getting started out launching a campaign and you’re just waiting for something, some sort of signs of life … And then you see something, you’re like, “Well hey, that’s something we can play with.” Resist the urge to make changes to it right away, because it’s just really common and it can, again, kind of screw up the account.

Segmentation + Statistics

So I think that something that is going to be useful in understanding just the best way to approach segmentation is just a little bit of statistics background; like we don’t have to be testing experts in AdWords, in paid search, it’s a different field, right? It’s a different discipline. But we have to at least understand the basics of statistics, how it works and what it means. Because we need to understand the concept of statistical significance, which means that there’s basically an association between two variables or two different metrics that we have that’s not likely to be due to chance. So that’s what we’re really going for; is this thing happening just due to random chance, or is there some reason that it’s happening? That’s what we want to understand. And when there’s a reason, we can unpack the reason, and we can optimize for that reason, but if it’s just well, a conversion happened to come in at 2 AM and a conversion happened to come in at 4 PM, there’s not a lot that we can do with that. We can bump the bids then, but there’s not a lot that’s gonna happen for that.
So the best way that we’re gonna reach statistical significance is getting as much data as possible.

Now I don’t optimize based solely on statistical significance, especially when we’re talking about conversions, and we may be limited in conversions and limited in budget. It can make sense to make decisions really just based on performance. And are there elements of chance? Yes there are, but if I launch two ads and one of them, let’s say it gets a hundred clicks, and one of them drives 20 conversions, and one of them drives one, that tells me that probably the first ad is gonna be better for the audience. Like I don’t need to wait for it to hit statistical significance to make a decision on that….

(Statistical Significance calculator from AB Testguide) 

But it is something that I’m going to be looking at, and again, this is where a lot of newbies, if you’re brand new to paid search, it might just be really tempting to say, “Well a 6% conversion rate is better than a 4% conversion rate, so six is the winner” without waiting long enough for the data to come in to really tell that story.

The other thing to keep in mind here is that AdWords is not a true testing engine. AdWords is motivated by profit, by its desire to get profit, and so what it’s gonna do is that once it launches things, it’s not completely random; it’s not doing an A group and a B group totally randomly. It’s choosing to serve or not serve your ads based on what it already knows about your ads and the history of it. So what will happen is that you’ll launch something new and most likely, it’s not gonna do as well as a thing that’s been launched for awhile. And after a few weeks, most often that will start to change, and AdWords will get more data and say, “Hey, these click-through rates really are better. Let’s start to serve it more” as it kind of gains that trust that your new ad is gonna perform better.

But it can take a little while, and I know that I’ve spoken with reps from AdWords and Bing who have talked about having clients that will launch something, see it’s not working as well, and turn it off after just a day because they’re like, “Oh, that was a fail.” They don’t want to waste more money. But a lot of times you do have to continue to invest in something for long enough to really see if it is working, and typically if you’re launching something new, it’ll just be behind the thing that’s existed for a long time, because that’s already proven, and your new thing isn’t proven.

You can’t just launch something new and just tell AdWords, Hey, treat these equally!

So I hope that just makes sense as a concept, that you can’t just launch something new and just tell AdWords “Hey, treat these equally.” They’re not gonna to do that; they’re going to treat it based on what it already knows, and if it doesn’t know as much, it’s gonna be a little bit more cautious about serving it because it’s not quite sure if it can make money from it. This is a really important concept that will really affect everything that you’re doing in AdWords. Understand that AdWords is waiting to know more and to understand how something’s gonna perform before it just starts serving it a lot, or before it starts to give it certain preferences… before it starts to say, “Okay, well this is very likely to get clicked, we’re likely to get money from it, and so we’ll go ahead and basically give you that discount on how much you’re paying for your CPC to reward it.” So that’s basically the landscape of testing. So we’re not gonna be running true tests within AdWords; we’re just going to be doing what we can.

Variance

So one other thing to speak to on this is variance. If you were to launch the exact same ad and just launch the same ad twice, it’s very, very likely that one is gonna out-perform the other. And as it continues to out-perform the other, based on what I was just talking about, based on the AdWords algorithm and its desire to say, “Hey this worked once; let’s see if it works again,” and continue to kind of bet on that ad, the second ad is gonna become a little bit more anemic, even though they’re the exact same text. So you have to understand that that’s part of the game too. Like you can have things on rotate evenly, you can do whatever you want to try to account for it, but ultimately AdWords is trying to make money.

So it’s not just going to act as a testing engine. So again, just something really important to understand. I mean, that’s testing and statistics, and it’s not exactly segmentation, but when you’re gonna start segmenting things out and making decisions, that’s important context to have before you can start making those decisions, is just what is AdWords trying to do, and it’s not just trying to give you blind test results. It’s trying to … I don’t want to say rig the system, but it’s trying to get clicks, and it’s gonna get clicks by doing what it knows is gonna work already. Just keep that in mind.

Now most of the segments we’re going to have in AdWords are automatic. Some of them have to be added and I’ll talk about that in a minute, but especially if you compare this to really any other platform, especially a lot of the paid social platforms that you’ll find, AdWords is a lot more straightforward and you can learn a lot more from the data. So for instance, you can set your age range and come back and look at it, and see which groups from that age range performed differently. So if your age group is 18 through 65, you can come back and then start measuring that and say, “Oh, the 18 year olds really didn’t buy at all, and the 65 year olds bought a lot, so let’s exclude one of these and increase the bid on another.” You can start to make those decisions in a way that you really can’t do with a lot of other tools. And so it’s one of the really nice things about AdWords, is you get so much data. Sometimes you’re gonna kind of drown in the data, but it’s there for you as you want to use it. Only a few segments, at this point, really need to be added.

Account Structure

So with that, let’s look at account structure, because this is an area of segmentation that you’re doing yourself; you’re creating what campaigns you want, and what ad groups are contained within those campaigns, what ads and what keywords, and all the settings within each campaign. And so this part is on you, to make sure that you’ve done as well as you can. If you are bidding on a brand, or allowing brand terms into your account, by all means have a brand campaign, where you segment out all of the brand traffic into a single campaign, or multiple campaigns if you need to, but make sure you distinguish brand from non-brand. And that not only do you set up the campaigns, so you’ve got branded and non-branded, but that you also do your due diligence to weed out or negative match any potential brand traffic that could come through other areas and muddy your results.

Just to give you a quick example of this, I had a client who refused to bid on a brand, and yet they had one random campaign where they had not negative matched out the brand, and that campaign was doing really, really well because of some broad matches where brand traffic was allowed to come through. And so their interpretation was like, “Wow, these products are really great that we have in this campaign,” but the truth of it was, brand was … Life finds a way, and brand had gotten through to that campaign, which was making it the source of most of the conversions and revenue that it was seeing. So just be sure you’re very, very consistent in how you’re segmenting those out.

But you’re segmenting by performance, and typically, the amount of money you’ve invested in brands that people who are looking for are much more likely to take action than people who you could say are further … You know, at an earlier stage of the funnel, who aren’t familiar with your brand. They’re a little bit less likely to take action on your offer, so that’s just something to keep in mind.

Hyper-Segmentation

Another thing that I kind of mentioned before but I think we can dig into more here, is that hyper-segmentation really comes at a cost. We don’t want to put all of our keywords into just one campaign; we want to split things out and be able to optimize. But by going too far the other way, where a lot of people call this a SKAG, or a single keyword ad group, there are people who this is the only way that they’ll do an account. And I don’t mean to criticize that; if that’s what works for you, by all means do it, but it’s not what works best for the majority of accounts that I’ve seen or at least with the way I do it, because what happens is we’re …
Ad Words loves volume, right? AdWords wants to make the most money it can from each ad and each keyword. And when we take those ads and we do too much segmentation, then instead of having one instance of the ad with say a hundred clicks, just as example, so AdWords has learned about that instance of the ad a hundred times, and so is pretty confident that it can put this in an auction and it’s gonna do well. While instead, we have let’s say 10 different ad groups, or 10 different instances of that ad within each ad group, and each one only has 10 times; that’s less proven. A hundred is much more than 10, and so what’s gonna happen is that ad is not gonna be served as much, and it’s not gonna have as strong of, basically, a history or a score for AdWords to optimize for.

So we do lose something when everything behaves individually, and I think, to give you an analogy of this, you can think of it kind of like forming a union, or a co-op, or basically just leveraging the volume, even of people as they come together. A co-op or a union is much stronger than an individual alone. And having ads and keywords and ad groups that have a lot of volume is gonna be much stronger than breaking it out a million times across having it a bunch of different instances, if that makes sense.
Plurals is a really common example of this. I know that there are some cases where it might make sense to have plural or a misspell actually bid on separately, but for the most part, there’s not gonna be value of bidding on a misspell and having a separate ad for that, because you’re not gonna misspell your brand name, you’re not gonna misspell the most important keywords in your ad, and so just even that relevancy isn’t gonna be there in the same way. There’s just not much of a reason to do that. It’s something to keep in mind, and I don’t mean to ruffle any feathers, and I know some people really swear by SKAGs, but I would really encourage you to segment it only as much as you need to. And then from the segments that you have, you could have several keywords in your ad group, as long as they all relate very well to your ad, and they relate well to the landing page. But if your keywords are similar, there’s no real reason to break those up and to break up the ads across multiple ad groups when you don’t need to, so again, just something to keep in mind.

So all that is to speak to how you can segment your campaigns out to get better performance or to set budget caps and optimize based on how you have the campaign structure. Another thing you can do to segment within your account is actually use the segmentation feature that’s built into the AdWords interface. Now if you’re using the … It’s called the old interface, there’s actually a button that says segment, and eventually that’s going to go away in the new, which is currently the beta interface for AdWords; what you’ll currently see is basically, essentially a pie chart next to the filter that is segmentation. So it doesn’t say the word segment, but it has the same segment features. And when you access this, then it gives you the option to segment your account performance in different ways.

Now the options that are available to you really will depend on what level of your account you’re in. So what you can see when you’re segmenting a campaign is gonna be different than what you see when you’re segmenting keywords, is different than something else that you may see, and it may not even appear for certain levels of the account. But what you can do … And I’m just gonna go ahead and read you what is available right now, because this is something that is subject to change, and so it may not be available in the future. But right now, you can segment by time; you can segment by day, week, month, quarter, year, and day of week. So that all, I believe, is pretty straightforward. Day of week is gonna be, again, the classic Thursday compared to Friday, and everything else is just gonna show you basically that timeframe. This is really obvious. I’m not contributing anything by reading it.
Click type, this one’s pretty interesting actually, because it’s gonna tell you where the clicks came from, whether it was someone clicking on the headline ad, or someone clicking … Particularly when you’ve got location extensions enabled, there’s gonna be a lot of different areas you can get clicks. So if someone’s clicking on different extensions or in different places on your ad, this gives you clues about how they’re interacting with your ad.

Segmenting By Conversions

Conversions is gonna show you conversion action, conversion category, or conversion source. I segment this a lot by conversion action so I can see, just at a high level, how my different conversion types are doing within an account. Or excuse me, within a campaign. And something that’s important to keep in mind for this is that there’s no good way to control the actual cost per conversion, or cost per action, when you start to segment. And so what it’s gonna do is it’s gonna take your total spend and divide it by if just this particular action led to all those conversions. I hope that makes sense; it’s a little bit funny, but if there’s multiple conversion types on a page, there’s no real way to say how much was spent driving each of those individually. So it’s just kind of collectively, and so to get to that amount, it’s gonna be broken out just as if there was that conversion. So don’t freak out if you do this and you see that your cost per conversions are really high for each of those, because that’s not a super meaningful amount. That’s just how it’s being calculated, so don’t worry about that.

Segmenting By Device

Device you can segment, again by … This is gonna be your tablets, your desktop or computers, and your mobile. So this is a nice way to segment because you can see how everything’s doing, and again, this will be by campaign or whatever level you’re looking at. And so this is gonna help you make decisions about how to optimize based on how people are reaching and interacting with your content. A lot of times right now, what we mostly see is that tablets don’t do that well, and mobile tends to have a lower cost per click, and lower conversion rates. Now this is just the most common scenario; it may not be the case for you, it’s not always the case for me. I have a client where clients actually out-perform everything, which is crazy to me, because they usually do so poorly. So you’re gonna want to check, but if you see that for instance tablets are doing horribly, you can reasonably bid that down, because that’s … There is meaning to that because of the device type. And so you can adjust based on that.

Segmenting By Network

You can segment for network and search partner, so if for instance you are opted into Google’s partner network and you see that it’s not doing nearly as well as you would expect, you can turn that off. You can’t currently opt in to just be part of the partner network without being on Google itself, and so if that’s performing much better for you, you’re kind of out of luck; you still have to be on Google as well. But you can do it the other way if you see that that’s not working for you.

Segmenting By Top VS Other

And then top versus other, this is the final standard segment, just again worth mentioning to give you some ideas about what you can do within the interface. Top versus other is gonna show you if it’s … Well, it’s a little bit archaic. It doesn’t make quite as much sense now because we don’t have the side, but you can still be below the organic listings, and so it’ll tell you if your ad was on the top or in some other place, and sometimes there are ads that still show up on the side, just not text ads. So it can give you a little bit more information about if you want to maybe adjust your bid, you can see the performance of that.

Adding Your Own Segments

Now the other thing that, for segments, I’ve told you that most segments are built in and so clicking on segmentation is certainly built in for you. But there are a few segments you can add. For locations, you can basically go to location, add different locations that you want to target, and you don’t have to adjust how much you want to spend on it; you don’t have to adjust the bid, or the bid settings. You can just add those segments to your account, basically to observe, and if you see that it performs differently, then you can start to make a change for it. And that’s also true for something like household income, which you’d do from locations.
And then within audiences, there’s this terribly confusing way to segment that you can add to your audiences, which is called target and bid, and bid only. And thankfully, this is going away, because honestly I get confused by it all the time; it’s just not set up in a way that makes any sort of sense to me. Like target and bid, just the way that that is structured and explained, I get confused every time I deal with it. But in the new, in the beta …

Currently the beta version of the AdWords interface, you can see that there’s ways to narrow your targeting, and this is kind of like the “and” feature of a Venn diagram, or saying all these things have to be true, or it’s called ad observations, which is essentially what bid only was. And so you’re just adding different segments, and you’re saying “I’m not gonna touch it; I just want to see what’s happening,” so just, you know, be on your way, but I want that extra level of insight or of data. And then if you do see any sort of insights that you would want to take action on, you can take action on it at any point you want. You can adjust the bids, but you’re not doing an “and” feature. So hopefully that makes sense.

Changing Terms

So target and bid is now gonna be called narrow targeting, and bid only is ad observations, which I do think is a little bit more clear than how it was set up before. So just kinda keep that in mind, that it’s going that direction and that is a welcome change.
Now I know it’s a little bit funny to be talking about what’s in the interface on a podcast where you don’t have any visuals, and I will be adding some visuals to the show notes, which again you can find at PaidSearchMagic.com/14. But really what I wanted to do is just help you get a better sense of segmentation. I feel like it can be really intimidating, when we’re talking about segmentation, well what exactly do we mean? But just keep in mind, it’s a way to isolate or slice and dice your data so you can better understand it and take action on it.

The other thing I want to remind you of is just hyper-segmentation isn’t always the answer. Like getting really, really specific and really, really granular in ways that kind of dilutes the power of what you’re trying to accomplish, that dilutes the volume, dilutes what AdWords is able to learn about a specific instance of your creative assets, that’s not good. So give it as much volume as you possibly can without doing harm. So for instance, I wouldn’t have a branded keyword and a non-branded keyword in the same ad group, but I would have my brand and a plural of my brand, or some other maybe modified version of my brand in the same ad group if it’s functionally the same; if there’s no real difference between how they perform, or what would inform a good ad for it, or what would go on the landing page. I would keep those things together so that AdWords gets as much data and information about my ad and my keywords as possible. I don’t like to unnecessarily segment things.

So keep that in mind, and also again, just be really mindful of how much information you have before you start to take action, and go in there, and make edits. Because it can always look like you have a winner and a loser, because you’re always gonna have numbers to compare to each other, and you can always say, “Well 1.15% click-through rate is better than 1.05% click-through rate.” So winner and loser. But a lot of that is just noise, and it’s just random. And you’re gonna have someone who’s really, really motivated and excited to take action on your offer, who will click even if it was your worst ad ever, because they’re so excited about something that they saw, whereas someone else could have the opposite situation; that they weren’t going to take action no matter no what, and your best ad’s in front of them, but they just are not interested at all.

So keep that in mind. There’s a lot of variation that’s going to happen and so when you’re looking at your segments, be mindful of the natural variants that’s happening, and try to only declare winners, or only take action, on something that approaches statistical significance; it approaches a state where you’re confident that it’s not just due to chance. That’s really the main idea here. So we were able to take action on things, and we’re taking meaningful action.

So that is what I wanted to talk to you about today for segmentation. I hope it clarified things and didn’t add confusion, but if it did, if you have questions about what this was, you can email me at amy@paidsearchmagic.com. Please don’t email me just to tell me I’m wrong about SKAGs, because I know that there’s difference of opinion, and I think at a certain point, this is just chef’s choice where I’m gonna say I don’t think, in the majority of cases, there’s a real need for it. You might disagree and that’s fine, as long as your account is working as you’re comfortable with, that’s wonderful and that’s what matters.

One other thing I wanted to mention is that I’ve done my best to make this accessible for you, but I know it can still feel complicated, and you can still say, “Okay, well that opened up more questions than answers,” in which case…

Learn more about working with Amy here.

Podcast: Play in new window | Download (Duration: 40:53 — 74.8MB) | Embed

Subscribe: RSS