Here’s Why You Should Separate Match Types By Ad Group

By , Owner of ZATO

Leaderboard_flashdoubleclick

ppc match type debate meme

 

If you pay any attention to PPCChat, you probably saw a good debate going on over there last Tuesday: Ad Group Creation Strategies.

 

Debates are healthy because they remind us we don’t have all the answers and that other people have great ways of looking at things. This chat was revealing to me, because I had always thought separating ad groups by match type was more mainstream than it is. It was insightful to get some new thinking on this topic to chew through my own reasoning. The reason blindspots scare me, is because they are blindspots! Where discussions and debate can be invaluable to us is that they help strip away our blindspots. Part of me is terrified to throw my thoughts and analysis out there, but on the other hand I recognize I won’t grow without you smart people picking apart my logic and thought processes. Would you join me in having civil and continued discussion on this topic?

 

That being said, I think I have a case for why it’s still worth it to break out match types by ad group… BUT… I also admit that I’m limited by the inner workings of my brain. So I would value your input and thoughts in the comments and on twitter: @PPCKirk. I fully admit that the majority of my post rests on heavy use of Adwords Editor. If Editor is not something you are familiar with, I would suggest learning it well since it is a significant source of efficiency for the PPC Manager (read to the end and I will give you insight into how you can use Editor to bulk copy/add separate match type ad groups)!

 

I’m not a huge fan of the “response” article, but I think this is a time when it can be useful.

 

Therefore, I will analyze 4 opposition arguments and explain my reasoning for dismissing them. It should be noted that the majority of my arguments relate to new account builds. I acknowledge taking over an existing account can get messy and it would be unwise to always (or even most of the time) blow it up and kill account history by changing all ad groups over to separated match types. This article will primarily present arguments for creating a new account.

 

4 Arguments Against Breaking Out Ad Groups Into Match Type

  1. Efficiency in Account Creation
  2. Efficiency in Ongoing Account Management
  3. Intent is Same, So Why Separate?
  4. Waste of Time Until Volume is Established

Argument 1: Efficiency in Account Creation

 

This is an interesting argument since it appears at surface level to hold validity. “In order to create separate Broad Match Modified (BM) and E ad groups, you are in essence DOUBLING the number of ad groups you have to create!”

 

Makes sense, right?

 

However, there is faulty reasoning here. Can you spot it? The faulty reasoning is: “Double the ad groups = double the time in account creation.” I was skeptical about this statement because frankly, using Editor, I can blast out multiple ad groups pretty quickly.

 

So for fun, I timed it. I went into a real client account, and made a test. Using similar keyword modifiers so I was as much as possible comparing apples to apples, I created 1 ad group with both exact and broad modified match types. I then created 2 ad groups, for another similar keyword phrase. 1 ad group was exact match. 1 ad group was broad modified.

 

These ad groups included:

 

– 2-8 KWs
– A/B Ad Test set up (copied ads from other ad groups to all AGs)
– mirrored Negative KWs set up in separated MT ad group

 

THE RESULT

 

1 Ad Group, Both Match Types: 2:15 (135 seconds) From start to account push up to UI.
2 Ad Groups, Separated Match Types: 3:15 (195 seconds) From start to account push up to UI. (actually I ran this twice, one time it took me 3:30 and the other time it took me 3:00 so I averaged this at 3:15).

 

What I learned was that you actually have to use a similar process to create BM and E keywords, whether in the same ad group or separated so the time saved by grouping them is not as significant as first appears.

 

Ad group, ads, everything else can be copied over to create the duplicate ad groups. The real time comes from adding the exact KWs as mirrored negatives, and even that is pretty easy with the Editor bulk operation.

 

adwords editor bulk negative keyword

 

Now, I understand this isn’t super-duper exact science but you do have to start somewhere, and even if things vary, or some ad groups take you longer/shorter than others, I believe the relationship between the 2 stands. It does not take “double” the time to create an account by separating match types out.

 

Hourly Estimate For Separating Match Types

 

Let’s put an estimate on it. Stick with me here because this comes in handy in our next point.

 

At my numbers, it will take approximately 30% longer to create the Search campaigns in your account. This doesn’t take into account any of the account setup for Shopping, Remarketing, Display, or account creation. For sake of argument, let’s say it takes 20% longer in an account to separate match types. In a 20 hour account build, this will add an additional 4 hours on the beginning. In a 15 hr build, an extra 3 hours, so on. Again, these numbers can get fuzzy but for the sake of this argument, I do want us to make an attempt to get real numbers so we’re not just dealing with gut.

 

So WHY the heck go to all of this trouble, Kirk??? Because, now your time wasted on separating match types is DONE. Over. Ended. Quantified.

 

On the other hand, the match type grouping model affects account efficiency for all time and into eternity. How? Keep reading…

 

Argument 2: Efficiency in Ongoing Account Management

 

The argument for this generally goes something like this: “Ongoing management needs to be efficient and organized. So many unnecessary ad groups creates account waste and visual confusion. It takes more time to manage an account with double the Search ad groups from match type separation.”

 

Here again, is another argument that sounds great at face value. Who would argue with decreased efficiency?! “That idiot Kirk, there he goes again wanting more confusion in his accounts. *eyes roll*”

 

However, I would push back on this one by again trying to dig a little deeper.

 

First, I’m not sure I understand how this decreases efficiency. – When I analyze an account, I usually sort by my top cost/traffic/profitable/revenue/whatever ad groups, keywords, queries, you name it. In other words, I am focused on those top performing segments of data, not on the aggregate whole. Who cares if there are 500 ad groups now in the account rather than 250, I’m still looking at the same data to make decisions as someone with match types grouped.

 

ugly llama picture break

 

Second, I will argue that grouped match types is actually more inefficient when pausing more specific match types in ongoing account management.

 

Example:
You are digging through your SQR and there is an exact match term that has been bugging you for awhile. You have been watching it, and are finally ready to kill it off. The problem is, if you have bundled match types in an ad group, you can’t simply pause the exact match keyword right in the SQR because it will continue to show up for the BM or phrase match terms in that same ad group. So what do you do? Well, now you have to jump out of SQR and into that ad group and then choose to add the exact match as a negative. Additional, ongoing steps from now and into eternity for what should have been a simple task.

 

If your ad groups were separated by match type with mirrored negatives, you would have already set these two apart from the beginning. Ready to pause thatexact match term in the SQR? Do it and move on with life. Your negative KWs bulk added in the beginning (according to my Adwords Editor step-by-step tips below) will automatically filter this one out.

 

Third, I will argue that it is actually more inefficient in the long-term to group by match type because of bidding inconsistencies. – When keyword match types are grouped together something called tiered/stacked bidding is necessary to increase account efficiency, and according to this Wordstream post, even profitability: Bid Stacking in Adwords: How to Pay Less When Using Multiple Match Types (see also Justin Fried’s post on Tiered Bidding: How to Implement a Tiered Bidding Strategy).

 

This, then, is the single biggest issue I see with grouping match types (for those without bid management platforms). When you use the sorting/filtering method to target individual top cost/traffic/converting keywords or search queries you are making decisions pulled out of the context of the Ad Group in which it resides. This almost always results in a failure to keep up with tiered bidding. I have never taken over an account with grouped match types where the bidding was not, frankly, a mess.

 

So how does this increase inefficiency you may ask? I’m glad you asked. In order to actually keep up with tiered bidding, you must make a choice EVERY TIME you bid up a keyword. You must jump out of your SQR, navigate to that ad group, and spend more time now analyzing what to do about the other match types in that ad group.

 

Yup, I counted that time as well 🙂

 

20-45 seconds is what it will take you to do this for each keyword you do an SQR to ensure tiered bidding is being followed.

 

So, if you spend 1 hour in SQR each week and if you are following the best practices of tiered bidding, you are decreasing your efficiency significantly. In fact, if you do it as well as you could (hoping into the ad group, analyzing how a bid adjustment on a BM term will affect the phrase or exact matches in that ad group), you are spending an extra 15-30 seconds per keyword with this approach. But, that’s minuscule, right?

 

Let’s take an easy account and be the generous with the overages (so against my case). 25 KW SQRs analyzed weekly at an extra 15 seconds per KW. That’s an extra 6 minutes per week. That’s an extra 26 minutes per month. That’s an extra 5 hours per year (you just passed your 1-time extra build time-overage if you had separated these from the beginning). Have the same client for 3 years? You spend an extra 15 hours just in minimal SQR work on that client and that number never stops rolling.

 

Are these extra steps for the bundled match type ad groups the end of the world? No, but I do think it negates the “efficiency” argument.

 

The Bundled Match Type Ad Group Choice: Increase Inefficiency or Reduce Profitability

 

The funny thing is that from my point of view, this is (probably unconsciously) known by most PPCers because they choose to go with Option B. They just ignore tiered bidding in the account and let whatever keyword get whatever bid they deem necessary in an account. But is that really the best for the account? If there is any truth behind the articles I shared before (which I agree with), than no. It’s not best for the account.

 

In summary, a grouped match types proponent must choose 1 of 3 less than ideal options:

 

  1. Keep up with best practices and implement tiered bidding during each SQR decision (thus significantly decreasing ongoing account efficiency).
  2. Ignore tiered bidding & save account efficiency, but mess with SQR as exact/phrase matches get pulled into higher bid broad modified and thus reduce account profitability.
  3. Utilize an automated bidding platform or system to keep up with this. @Mel66 suggested this on Twitter (and it’s a good idea for those tools that work well!). Of course, this adds additional cost into the equation since these are not free.

 

On the other hand, as demonstrated above, I believe the only true additional time a separate match typer will add on is the brief setup time (roughly 3-5 hours). Ongoing SQR optimization can be managed happily and quickly since there is no danger of BM cannibalizing the exact queries.

 

Argument 3: Intent is Same, So Why Separate?

 

I’ve not heard this argument as much, but I wanted to address it briefly since it has been brought up before. Basically the argument is that if you choose your keywords well, the user intent will be the same so there is no need to segment out into separate ad groups.

 

I don’t have intense verbal arguments here so I will revert to an intelligent, timely, and compelling meme:

 

britta perry logical argument

 

I’ve found that broad match modified nearly always sneak crazy intent queries into the mix. This is the main reason the SQR is so valuable. It helps us identify keywords so far off that we need to exclude them, or keywords separate enough that we need to build out new ad groups.

 

Frankly, I go the opposite on this one. I think user intent is so separate between BM and exact terms that it aids the argument to break them into separate ad groups.

 

Argument 4: It is a Waste of Time Until Volume is Established

 

This argument is similar to the “efficiency” arguments, but I wanted to pull it out and analyze the volume aspect further. Of all the arguments, I think this has the most merit. Basically, the proponents of this argument are agreeing. “Yes, splitting match types out by ad group is the way to go, but wait until you know which keywords/AGs to focus on so you aren’t wasting time.

 

I completely agree that I don’t want to spend time copying/pasting that I don’t need to spend. However, I don’t think this argument is as simple as that upon closer examination.

 

I think the 2 key problems with this argument are:

 

  1. It is less efficient. It takes more steps & touches to analyze (identify, open editor, copy/paste for every individual ad group you want to change from now until forevermore) & then create a new ad group off of existing grouped match types then to simply bulk create all separated ad groups in the beginning.
  2. It causes all historical data to be lost for whatever match type is being moved. This kind of speaks for itself, if you have a high volume ad group you want to break out you have to choose which of those match types to sacrifice your historical data on the altar on optimization.

 

Perhaps this will reveal different account building strategies, but I’m not doing a full analysis on every keyword I’m putting into my accounts in account creation. Honestly, I think THAT is a waste of time since there is no negative to this strategy (not getting volume after a month, just pause it and move on). I analyze the main keyword groupings and develop keyword modifiers and themes based upon them so I can rapidly bulk-create ad groups off of those originals. We’ve all been surprised by keywords that demonstrate more/less volume or conversions than we expected. So going off of the presupposition that adding the additional match type breakouts in Search campaigns adds minimal hours of extra work to the creation, why would we want to create a full account 1 way, and then “optimize” it by completely changing it around in an ongoing capacity in the future?

 

Conclusion

 

So, have I convinced you? Am I full of it? Have I merely hardened you by demonstrating my lack of knowledge in one of these areas? I want to hear your thoughts below and on Twitter! Debate and discussion is how we learn. PPC is an evolving and complicated game so I will be the first to admit I have room to grow. Help keep the conversation going by sharing your ideas on this as well.

 

BONUS: Adwords Editor Step-by-Step

 

Let me close by sharing a trick I like to use in Adwords Editor for bulk creating separate Ad Groups (no spreadsheet exporting required). Perhaps there is an easier way to do all this, and if so please share but this is as easy a way as I’ve found so far!

Step By Step Bulk Edit Separation of Match Types in Adwords Editor

Step 1 > Create All Exact Match Type Ad Groups (with keywords & ads), make sure to include an identifier in each. I use (E).
Step 2 > Copy All Ad Groups & Paste.
Step 3 > Highlight New Ad Groups. Find/Replace (E) with (BM).
Step 4 > Filter for all keywords in Ad Groups With (BM). Now only keywords in your BM ad groups are showing.
Step 5 > Highlight all KWs in your view and bulk change from exact to broad. (You can also bulk adjust bids slightly lower now if you want).
Step 6 > Still highlighted, Find/Replace again. Find all spaces ” ” and Replace With ” +”. Note, there is a space in front of that +. This will add the + Modifier before all unique words in these KWs except the first word of every KW.
Step 7 > Still highlighted, Find/replace again. Append “+” to the front of every KW.
Step 8 > Filter for all keywords in Ad Groups With (E). Now only keywords in your E ad groups are showing.
Step 9 > Highlight all Exact Kws. Copy.
Step 10 > Go to the Negative KWs tab, select “Make Multiple Changes”, Paste all the E KWs you copied here. Select All BM Ad Groups. Apply.
Step 11 > Check & Upload changes to UI.

Twitter Facebook LinkedIn Google+ Email Print More

42 thoughts on “Here’s Why You Should Separate Match Types By Ad Group

  1. AnnieintheSun

    Kirk, nice post to share the rationale on Ad grouping. I’m just starting to get down and dirty with this part of AdWords and had to smile when I came to the crazy llama face. I’m making the case for entertaining business content and this is a great example. Stuffy is gone; long live the interesting and novel! (Looking forward to your Thursday webinar.)

    Reply
    1. Kirk W

      Thanks Annie, I figure, if I’m going to make readers suffer through that length of post then I may as well try to get them to smile once in awhile 🙂

      Reply
  2. Faye

    Interesting one.

    I have never created an account from scratch, I’ve always inherited.

    In the accounts that I have managed they were always in one ad group to begin with then split them out. The number of exact match keywords never = broad match (there was also more broad). We then went through the SQR & pulled out terms to exact match when there was reason too. Eg high volume/conversion % etc.

    If I was to create an account from scratch then I would split it out my match type again.
    If you know your way around excel & editor then it wont take too much longer to create (as you pointed out)

    From what I understood the performance of keywords in the ad group effects the rest of the keywords in that ad group? So if a broad match has a lower QS then it would effect the QS of the Exact match?

    Reply
    1. Kirk W

      Thanks Faye!

      Confession, I’m not a Quality Score expert so if someone else weighs in on this I’d be interested to hear their input. My understanding is that since real (not to be confused with Visible) Quality Score does take into account multiple Match Types, that separating by Match Types could have an affect on helping real Quality Score.

      It’s funny, I actually thought of QS when writing this post but (as I just demonstrated) I haven’t thought this through more exhaustively and the post was already getting stupid long! The conclusion I came to was that QS could not be an argument against Splitting out Match Types so I ignored it since it wouldn’t have hurt my argument.

      Reply
    2. iConversing

      “So if a broad match has a lower QS then it would effect the QS of the Exact match?”

      As far as i understand, it doesnt. As google stated in their whitepaper “Settling the (Quality) Score”, it says “There is no such thing as ad group-level, campaign-level or account-level Quality Score”. QS of other keywords are also not part of the static QS calculation (expected CTR, ad relevance, landing page experience).

      Reply
  3. Evelyn Baek

    Hi Kirk,

    I’ve actually gone a step further and split out my broad match groups into their own campaigns so I could control their budget better. I agree with you – I am a fan of splitting out match types and it doesn’t take much more time with some Excel magic.

    Reply
    1. Faye

      To jump on this, I split out by campaign too & not bad ad group.
      Although I have phrase match ad groups in the broad campaigns as we don’t have that match phrase match keywords.

      Reply
    2. siliconallstar

      Totally agree with you on this one. We break everything out by campaign and find it to work great.

      Reply
  4. Gianpaolo LorussoGianpaolo Lorusso

    I could even agree that there is not big minus in splitting ad groups by match types, but, as long as you are bidding higher on closer matches, and they are not your “top keys”, I still cannot understand what is the real plus in splitting them.

    Reply
    1. Kirk W

      Hey Gianpaolo, thanks very much for taking the time to share your thoughts. From my perspective, your comment represents a good number of people who don’t separate by Match Type. What I would do is to try to get you to dig in further into this part of your comment:

      “as long as you are bidding higher on closer matches”

      I believe stating it like this does not give justice to the more complicated implementation of a bidding strategy that will keep “closer matches” bid higher. As I note in Argument #2 of my article, the only way you can do this well is to spend significantly more time on each KW bid decision pulling open your Ad Group, analyzing whether you can and should adjust all match type bids and then making the decision on multiple Match Types (what started with the analysis of one KW). If match types were separated, you only have to make a decision on the MT you are analyzing!

      In other words, I don’t think it’s as simple as you make it out to be in just “bidding higher on closer matches”. That is the very issue I’m trying to get us to think about more.

      One other problem I actually didn’t mention in my article, is when your broader match types (in the same ad group) actually have better conversion history than your closer match types in the same Ad Group. This is not at all theoretical as I have seen it many times in grouped MT accounts. Then you really have a quandary on your hands. Do you raise the “great performing” BMM kw bid and leave the lower-performing E bid alone, but then blow Tiered Bidding? This is never an issue in the Separated MT strategy.

      Does that help more or still not seeing the benefit?

      Reply
      1. Gianpaolo LorussoGianpaolo Lorusso

        Substantially we are saying the same thing.
        The difference is in the approach.
        I separate groups for match types when I am forced to because is the only way to improve performances (and it happens quite rarely), you do it systematically.
        The only way to say which structure would be the best should be to run an experiment with this only difference and the same ads, URL, extensions and so on.
        Have you ever done this test?
        All the rest is theory 😉

        Reply
        1. Kirk W

          I agree and disagree. I did address the “volume approach” argument you reference at the end of the article (Argument 4). As I noted there, the inherent problem I see with waiting until volume is established is that you blow apart your history since you are taking existing keywords/ads and breaking them into new ad groups.

          I would push back that it’s not simply theory since my arguments have primarily been focused on efficiency and since I have attempted to identify specific ways in which efficiency is measured and increased by breaking out into ad groups (in this instance, basically time for creation and management).

          I would agree with you if my argument was primarily one of increased profitability (though even then, the call for case studies is not always helpful since accounts/industries react so differently because of various factors), but I have been speaking more to the efficiency aspect in which I have included specific instances of “better than”. I believe those are the test you were asking about which go beyond theory. The exact efficiency results may perhaps vary from advertiser to advertiser, but they did go past theory.

          Reply
      2. iConversing

        I dont quite understand the following:

        “Do you raise the “great performing” BMM kw bid and leave the lower-performing E bid alone, but then blow Tiered Bidding?”
        Yes of course. Tiered bidding is only a guideline, why hold onto it when data proves otherwise.

        “This is never an issue in the Separated MT strategy.”
        Why? Wouldnt we be changing bids too?

        Reply
        1. Kirk W

          Hi iConversing, thanks for your question.

          The problem with ignoring Tiered bidding with different match types in the same ad group is because then you risk the more specific match type (say, [exact]) keyword queries being included in the (now) higher bid, but more general match type keywords (say, +broad +modified).

          So in essence, without Tiered bidding (or separate ad groups for match types), you are forcing all match types into that high bid broad modified keyword and kind of negating the reason for splitting them out in the first place.

          Reply
          1. iConversing

            Thanks for your reply.
            Er do you mean your manual bidding is at the ad group level?
            We do it at the keyword level, so it is ok to up bid the BMM because as you mentioned, sometimes exact dont convert as well as the BMM.
            Am i missing something?

          2. Kirk W

            No, I mean KW level bidding as well. If you bid the BMM up higher than the exact match, than you are sending some/all of you exact match clicks to the BMM. You don’t want that, you want to keep all your exact match queries going to your exact match KW, especially in this scenario when the exact is lower converting… otherwise why would you have broken it out into 2 keywords in the first place? Therefore, the only way you can ensure the lower-bid exact match keyword does not appear in the higher bid BMM KW is by using Tiered bidding or breaking them out into new ad groups.

          3. iConversing

            I see. To me, this benefit of negative keywords in the phrase/BMM ad groups is enough to justify the breaking up by match types. This actually works better than tiered bidding, because to me the adjustments of tiered bidding by % is rather arbitrary?

            One thing however, i think only in rare cases will exact matches be eclipsed by the BMM even if the exact match bid is lower resulting in lower Ad rank. Thats how its meant to work because exact match type has higher priority than ad rank for keywords grouped in a single ad group.

            The BMM (in the case above) can only outrank phrase matches, but not the exact matches. I definitely do not think phrase matches are dead, or that match types will be dead anytime soon either.

            So do you actually create 3 or 4 ad groups to separate by match types?

          4. Kirk W

            Yes, the negatives is why I prefer breaking them into separate ad groups from the beginning. I typically only bid on BMM and exact now, with phrase when it is a brand term or other high profit/easily confused term. In those instances, I will break them into 3 ad groups (BMM, P, E), but that is rare. I have also been experimenting more with Exact campaigns and BMM campaigns rather than ad groups so more budget can be devoted to the exact keywords. No perfect way, but it has worked well for me.

          5. iConversing

            Thanks for sharing. I will think it is ok to keep the exact and phrase match types together as the exact will have priority, so we get lesser ad groups.
            In order to keep the statistics, i prefer using flexible bidding strategies so we can assign budgets by ad groups without breaking them into campaigns. Anyway, whichever works!
            It was good discussing this with you, thanks.

  5. Catherine Wilson

    We always break out our adgroups by match type – thanks for confirming! Question though… you mention BM & E match types, but not Phrase. Do you not use phrase?

    Reply
  6. Catherine Wilson

    Thanks! We had an interesting discussion on the use of phrase match today based your article and agree with your response that it is still relevant. We will continue to use it since we have the bandwidth.

    Reply
  7. JVH

    I can’t say it seems worth the while for me. Working with a relatively large account (think tens of campaigns, hundreds of ad groups) I think I will lose oversight of which products are or are not performing well as they’re all split up. Also, phrase match is actually the best performing match type in this account, meaning I’d probably have to split that up as well.

    Reply
    1. Kirk W

      Thanks JVH for your comment, I’m not sure what to tell you other than that I work on a variety of sizes including the account size you referenced and still believe in this strategy.

      I think the assumption here is that it’s easier to keep track of top performers with less ad groups, however I think utilizing good filter practices in the Adwords UI helps you focus on the top performers/top problems quite easily regardless of the number of ad groups.

      Reply
  8. Cassie Allinger

    Hey Kirk — Great work. I love a great debate, and although I missed the #ppcchat for this one, I wish I had been there. I think you made some very solid arguments, but there’s one particular argument that you didn’t cover. Separating adgroups (or campaigns) by match type has proven to be an issue when you need to use Google’s conversion optimizer, as it dramatically decreases your volume at the ad group level. While volume won’t always be an issue, it often is. Most of the time, if I opt not to separate by match type (or to merge all match types) it’s for this reason. And for the sake of argument, let’s say that ditching conversion optimizer isn’t an option. Thoughts?

    Reply
    1. Kirk W

      Cassie, thanks for writing. Honestly, I didn’t think about the conversion optimizer impact, so great thinking. What you said makes sense, but am I correct in assuming that this really only affects a smaller number of accounts?

      Frankly, I’ve tried conversion optimizer and so rarely seen positive results, that I am not a fan of it. I know some people like it and I don’t doubt that it works for them. What I am aware of is that volume plays a significant role in conversion optimizer (as you alluded to), so perhaps for a large account that has the ability to be successful it is worth it. My big concern as I mentioned in the article is maintaining a Tiered bidding strategy, but in an account large enough to be successful for conversion optimizer it also stands to reason that they could afford some sort of bidding software to help manage tiered bidding.

      So I will begrudgingly admit, (1) if you can make conversion optimizer work for you and (2) if you see a significant drop in conversions from splitting match types (3) if you use a bidding software to help maintain tiered bidding thus preventing inefficiency and lost profit… then I could see this as an argument for merging match types. 🙂

      Reply
      1. Cassie Allinger

        Why thank you, for begrudgingly admitting, that I have a solid argument here. 🙂 I wouldn’t necessarily assume that this only affects a small number of accounts, although I don’t have any data. Anecdotally, it affects enough to be considered. I do agree with your tiered bidding argument, but it is not an issue IF you’re working with conversion optimizer.

        On another note, GCO isn’t my favorite tool either, but I have seen it work. Sometimes a situation will demand it. I believe it’s largely misunderstood and misused, hence the negative reputation. In my experience, most people try to apply traditional PPC optimization logic to GCO, and that’s where it fails. Your strategy must change to reflect how the tool works. Once I started to adjust how I approached working with GCO, I began to see its value in some cases.

        Reply
        1. Kirk W

          Yeah, I can see that as well (my begrudging comment was meant to tease :).

          My assumption of small number of accounts is taking into account the lower number I am aware of that do GCO well (I’m not placing myself into that group!), as well as the number that have the volume to do it well. My assumption is that where those two cross is a small number.

          On the other hand, I fully admit I could be wrong! Perhaps many more will come out of the woodwork admitting to it. That’s a negative of being somewhat siloed as an independent operator, I’m dependent on (1) my own limited client base, and (2) what people are willing to share on Twitter for my assumptions!

          Are you going to Hero Conf? I would LOVE to learn more from you on GCO.

          Again, thanks for your thoughts on this. Valuable information.

          Reply
          1. Cassie Allinger

            Of course! Unfortunately I won’t be at HeroConf this year, but happy to chat anytime.

  9. Vlad ElRuso Markovich

    Hi Kirk, imagine you have a big ecommerce account, with 50 campaigns, 500 adgroups. It’s probably easy enough to create these campaigns with exact kws separated. But what about the optimization? Do you think it is handy to add negatives to broad adgroups each time you add new kws to the exact adgroups? Or vice versa, delete the negatives when you want to pause a low performing kw? I do it weekly, I add about 100-300 new kws and pause 50.

    and more importantly, you are not saying a word about the advantages of this separation… but your article says: Here’s Why You Should Separate Match Types By Ad Group

    Reply
    1. Kirk W

      HI Vlad, thanks for the comment. It seems like you were unhappy with the post so I’d like to take some time to address your concerns in detail.

      2 initial thoughts. First, philosophically, I think your questions are focused on the “management time” aspect without also bringing in the “results” equation into the mix. If you can demonstrate in this account that separating match types is more profitable for the account, then yes, you should take the extra time for that extra optimization time since it is for the good of the account even if that means doing extra work on all the new kws/paused kws you are working on weekly.

      Obviously, that’s harder to do in an established account, so I can sympathize. You don’t want to pull the trigger on re-arranging things and risking messing things up if you can’t guarantee there will be better results. So I think you’re asking some great questions there.

      Practically, I agree with your concerns, especially in an established account. One of my current clients is an established e-com client the size of the example account you mention (larger, to be honest). I have not “blown” it up by implementing this strategy in full and completely changing everything over, but as I have gone through it and begun optimizing where it makes sense, I have implemented the match type segmentation strategy as I go.

      To answer your concern about my article title, I’m not sure if I agree with your opinion in this instance. Here are the points in the article where I specifically referred to the advantages of this separation. I chose to conceal the advantages within the addressing of the individual arguments, but they are there:

      – In addressing Argument 1: Efficiency in Account Creation, I began by addressing the argument against the MT segmentation, and then moved to demonstrate why/how one actually saved more time (the advantage) in new account creation by segmenting MTs.

      – In addressing Argument 2: Efficiency in Ongoing Account Management, I argued specifically why utilizing a MT strategy allows for a more accurate, and long-term Tiered bidding strategy (the advantage).

      – In addressing Argument 3: Intent is the Same, So Why Separate, I argued specifically that intent is different enough between Broad Modified and Exact match types, that it supports the separation of Match Types between ad groups (the advantage).

      – In addressing Argument 4: It is a Waste of Time Until Volume is Established, I argued specifically for 2 advantages to MT Segmentation: (1) that it saves time in quicker account creation then ongoing analysis and (2) MT Segmentation preserves historical data whereas waiting to segment until later can remove some of that data.

      Now, admittedly, I have continued to grow and learn since I wrote this, I would encourage you to check out the recent SMX Match Type Segmentation debate slides (I unfortunately was not able to be there) as they get into this more and I have much respect for the participants in the debate. I think there are some great points to be made on this as we continue to try to figure this out. If anything, if I rewrote this, I would likely add the word “consider” into the title: “Here’s Why You Should Consider Separating Ad Groups by Match Type.

      I’m sure we could find much agreement if we were to discuss this more, and I always hold to the notion that you need to do what’s best for your account so if it’s working for you, then keep on keeping on, but I hope I was able to address your concerns.

      Reply
  10. Marcus

    Hey Kirk

    We have done this in the past but I sometimes wonder what the benefit is over simply using tiered bidding in a single campaign.

    These figures are pulled out of my dainty behind but they give you the rough idea:

    [my keyword] – 100% bid
    “my keyword] – 75% bid
    +my +keyword – 50% bid
    my keyword – 25% bid

    We have found this approach to work well for most sites and we cover the spread and bid up where we have solid metrics against exact match. We still use the query report to add in negatives and more exact match / phrase match as time goes on.

    We have found a few clients where the exact match is very clear but it gets looser at phrase match and looser still at modified broad etc. When this is the case we have broken out to adgroups by match type as we can better target the ads and landing pages for any looser match. We still have to tier bids though and feed from these accounts into their own exact match ad groups.

    My thinking is that a tiered bid by match type is nearly always an banker but breaking out by match type depends somewhat on how the range of intent shifts from exact to phrase to modified to broad.

    Good read. 🙂

    Reply
    1. Kirk W

      Thanks Marcus, my concern as presented in the article is that without a 3rd party bid solution, it can be difficult to maintain that Tiered bidding structure longterm.

      I do admit, I don’t adhere to this as much as I did back when I wrote this in new accounts (since this is primarily about new builds, accounts or campaigns). If I am working with keywords that are going to have low volume, I’ll leave them together in AGs now due to the Low Search Volume label the exact match can get :/

      Thanks for the comment!

      Reply
  11. J.T. Smith

    I personally like SKAGs a little more that this method for high-volume terms. The reasoning is that if any search terms are not exact match variants, you know they are either irrelevant or belong in another ad group.

    If your ad groups are not tight enough in terms of relevancy, then how do you know an ad tweak is only doing better because one of your keywords are more relevant and another might have taken a hit.

    I noticed in some of the accounts I manage that the same search terms are being served in like 5 different places in the account. I think having the practice of adding every SKAG as a negative in all the other SKAGs makes sure that all identical searches are focused into a single ad group.

    Reply
    1. Kirk W

      I think SKAGs are a great option for those high volume keywords and this method can help more with overall account organization of the lower volume keywords. Since I’ve written this (and received oodles of feedback), I’ve definitely come to appreciate more of a pragmatic mixed bag in account organization.

      Reply
  12. Natanel Arnson

    I’m pretty new to PPC. I’m assuming the reason to split match types by ad group is to write ads that work for each match type? Specific ads for Exact Match, general ads for Broad Match?

    Reply
    1. Kirk W

      Hi Natanel, welcome to PPC! Ad testing is definitely part of the puzzle. I also like to separate match types because of account organization and better bidding control. Unless someone has a 3rd party bidding tool, keeping Exact and BMM separate allows you to keep them at individual bids without risking losing Exact traffic to a higher bid BMM term in the same ad group. There are lots of good arguments all around and this is definitely not the only way to do things so do some good research seeing what works best for you with account organization. Some people like to separate match types by campaign as well (to control budget). I have tried this on a few limited budget clients and like it since I can keep all Exact match keywords live with my limited budget. Single Keyword Ad Groups (SKAGs) are also very popular right now.

      Reply
  13. Nick

    Hi Kirk – Can you please elaborate on the following comment you made:
    “If I am working with keywords that are going to have low volume, I’ll
    leave them together in AGs now due to the Low Search Volume label the
    exact match can get”

    Forgive my ignorance, but why are you trying to avoid the “Low Search Volume” label. If the exact match version of that KW has such low volume that it’s not triggering ads then how would you have been able to change that (regardless of campaign structure). You still have the broad match equivalent that may pick up other BM variations and may reach enough volume on those variants to trigger the BM ad…but the additional volume from the broad matches wouldn’t have allowed the ad to serve for the lower volume exact match query, would it?

    Reply
    1. Kirk W

      Thanks Nick, I think you bring up a good question. In terms of how LSV keywords actually work in the auction, I believe your description is accurate.

      In my mind, the benefit to using a BM keyword instead of breaking out multiple LSV Exact match terms is that the BM term would be able to “cover” for ALL of those multiple LSV terms (as well as other queries) while those LSV terms go in and out of the auction. Therefore, the benefit is that because all data is funneling into this one BM keyword, you have enough for statistical significance and can make actual optimization decisions (bidding, etc) on it. Admittedly, leaving all of them individually broken out doesn’t change the actual ad auctions themselves, but it does keep the keywords so click-poor that you could never actually do anything to optimize them and just have an overly heavy account in terms of structure.

      Honestly, I could be thinking poorly here so interested to your thoughts as well.

      Reply
  14. Nick

    Hi Kirk – I’m not sure if you’re even responding to this thread any more but we did as you suggested and broke out all of our keywords into Broad, Phrase and Exact match campaigns. We did it at the campaign-level because we already have ad groups that we use to sub-categorize. The results are great. We were able to increase our bids on Exact and Phrase (because they perform so well) and we were able to decrease our bids on Broad to achieve a level CPA, but at a higher volume of clicks and leads. We can now see that our Exact and Phrase KWs were subsidizing the Broad but it wasn’t easy to see since everything was lumped together. Thanks for the tip!

    BUT HERE’S A QUESTION…Since Google has stated clearly that Quality Score is only assessed (up or down) on exact match impressions…then how does the “broad match only” campaign ever earn/improve quality score? In other words, Google typically expects you to run Broad, but still get exact impressions (where they can assess QS). In this case our BM campaigns never get a single exact match impression…so how in the heck is Quality Score assessed???

    Reply

Leave a Reply

Your email address will not be published. Required fields are marked *