The Ideal Number Of Keywords Per Ad Group

Share with:


A while back, one of our new hires asked a great question over IM about the number of keywords in an ad group. Here’s a paraphrased version of how the conversation went down:

New Hire: I’ve been told an ideal number of keywords in an ad group is around 15. If you have much more than 15, what are the chances all the words are relevant? Are smaller ad groups better, like in the 5 word range? Does it just make it more tedious to manage having a lot of small ad groups?

Melissa Mackey: Yeah, there comes a point of diminishing returns when you go below 10-15 keywords. That said, I’ve had 1-keyword ad groups for a very high volume term. It just depends – like a lot of things in PPC.

NH: Ok, so you look at diminishing returns and term popularity.

MM: Right, as a rule that works. Also, you might want to isolate smaller groups of keywords to improve quality score. So for example, if you have a few keywords with decent volume and poor quality score, you’d move them to try to improve it.

NH: What if you have a small ad group where one term gets impressions/clicks and the other one is extremely light?

MM: That’s usually ok as long as quality score is relatively similar.

Was I right about that? I’ll get to that in a second.

The “right” number of keywords in an ad group is a subject of much debate. I found a Quora thread that had as many different “right answers” as there were commenters in the thread.

Brad Geddes weighed in on the magic number of keywords in an ad group on the Certified Knowledge blog. Short answer? There is no magic number of keywords – it depends.

A poster on the Adwords Community forum does a good job of illustrating the concept, but then says 5-15 keywords is the right number.

I agree with him, to a point. I usually strive for no more than 15 keywords per ad group. But I also have ad groups with 50 keywords or more, and that’s fine too. It just depends.

The difference comes in whether the ad group is large because there is a large number of related terms out there, or whether the ad group is large simply due to laziness or lack of time. I recently did some keyword research around healthcare marketing, and there are a LOT of variations of “healthcare marketing” that are all closely related.

So how do you decide if you should split up a large ad group into smaller ones?

Look for similarities.

The first thing I do is look for similarities: in keyword theme, performance, or quality score. In fact, while I often say you shouldn’t optimize based on quality score alone, you can use it as a guide here in ad group development. Often the quality score will tell you what Google thinks is similar about the terms.

Quality score also helps you think about ad copy and landing page needs. If you have a bunch of relevant keywords with a low quality score and you’re not in an industry with traditionally low quality scores, then maybe your landing page isn’t relevant. Or maybe your ad copy needs to be tightened up. Creating new ad groups can be a way to deal with both issues.

Consider grouping by match type.

Sometimes it makes sense to group keywords by match type, to aid in keyword research and control cost per click by match type. I’ve found this especially effective for smaller accounts in niche markets where it’s hard to mine for new keywords simply by using search query reports. In larger accounts, grouping by match type just makes for unnecessary management time.

In fact, too many ad groups often become cumbersome to manage. Even a couple hundred ad groups can be super time consuming – I speak from experience on this. Single keyword ad groups (SKAGs) do make sense, but your entire account shouldn’t be full of them. You don’t want to end up in a situation like this:

twitter convo

This example is less about too many ad groups and more about an unreal number of negatives, but you get the point.

To me, the ideal number of keywords in an ad group is…. It depends. Surprise!

What’s your rule of thumb on number of keywords per ad group? Do you have a rule of thumb, or do you decide on the fly? Share in the comments!

Related Posts:

Comments

  1. Depends on how closely related the terms are to each other. You have Plurals and Singulars that usually belong in the same ad group [nyc apartments] and [nyc apartment]. Word order variations such as [AdWords PPC] and [PPC AdWords] that usually belong together. Exact and BMM versions (Phrase too if you’re into that sort of thing) that will often go together like +AdWords +PPC and [AdWords PPC].

    Then there are misspelled versions that could be accounted for like [accommodations] and [acommodations] and [accomodations] and [acomodations].

    There are also acronyms that may be viable… [PPC management] and [PPC MGNT]

    You might also want to account for variations with language and character differences such as [café] and [cafe].

    Or alternative spellings like [st paul mn] and [saint paul mn], or [twelve pack of beer] and [12 pack or beer] and [dozen pack or beer] and [case of beer].

    Or short, often unneeded versions that might hint at conversions differences such as [the rock] vs. [rock]. Or [homes nyc] and [homes in nyc]. Or [home sale] and [home for sale]. Or [tv free shipping] and [tv with free shipping].

    Don’t forget about spacing variations like [AdWords PPC] and [Ad Words PPC] and [AdWords P P C].

    So, yea. It depends. 🙂

    • Melissa Mackey says

      James, I love it! As you so brilliantly illustrated, there is an infinite number of ways to structure ad groups. I’d even argul with your first point that plurals and singulars belong together. When I worked in house for a magazine subscription agency, we saw totally different performance for [magazines] vs. [magazine]. In fact, each of these keywords had its own ad group in a VERY large PPC account due to the completely different performance. So yeah, it depends! 🙂

      • Great Magazine(s) example! I’ve seen similar patters for a clothing hanger client… Hanger vs Hangers. People often buy these in multiples like socks, shoes and gloves. Also the singular BMM version often matched to keywords related to other verticals/industries.

  2. Rule of thumb for me is not “what’s the right number” but how many is too few and how many is too many for me as and account manager *and* for this account/campaign/adgroup/keyword?

    In other words, some keywords must have their own ad group, and even campaign, whilst the majority can be in their semantically (and intention-related) themed ad groups with dedicated ads – but, of course, as you’ve mentioned, a “theme” is deliberately not a set number.

    The primary question is in 2 parts: Does this *keyword* (or keyword family) *need* their own ad copy – would keeping this list of keywords united in one ad group make semantic sense and share the same intent and would the same ads all do equal justice to it or should I split this up in some way? – and do *I* *want* to manage it (bids, match types) on its own or can it be lumped in with other ones?

    The “need” part of the question relates to the thought processes behind the decision – what is the volume and conversion performance of this keyword and also does it have unusual CPCs or does it perform significantly different on mobile vs computer/tablet. It’s similar to thinking “what sort of coffee table does this room need”.

    The “want” part is about time taken in managing the account – getting workflow efficiency so you’re not ferreting about in the weeds to manage some high impact keywords and so you can control where in the account the search traffic triggers ads. Going back to the room analogy, “Do I want to spend time building a coffee table from scratch after spending hours drawing blueprints or do I want to grab one from Ikea? Or do I want to get a pre-assembled one from Gardner-White?”

    A secondary question relates to the close variant feature on the main SEs: Do I even need to add this variant? What does it add to the account/ad group and does it perform differently from the current matching keyword? (“I have these occasional tables next to every seat in the room; do I even need a coffee table as well?”)

    • Melissa Mackey says

      Yes! And this, as you say Steve, is why companies need to hire a professional PPC manager. 🙂 Thanks for the awesome comment!

  3. I tend to have a slightly different focus when it comes to ad group size. I tend to ask myself two questions. 1) “Would I want a different user experience for a subset of the keywords in the ad group” and 2) “If I could break out the keywords and build a new user experience, how much of a lift would I need for it to be noticeable?”.

    If you would serve all the users the same ad copy, extension, and landing page, then I find it rarely helps to break down an ad group. I notice this a lot with misspellings of brand names. You wouldn’t do anything different regardless of how the user misspelled it, so why not group them all together (even if there are 50+ misspellings)?

    For the second question, if you find yourself with a smaller ad group where even doubling CTR for half of the keywords wouldn’t lead to a noticeable improvement, then you probably have better things to do with your time.

    • Melissa Mackey says

      It’s all about efficiency! You’re right that it often doesn’t make sense to split things up just for the sake of splitting. The only time I split out misspellings is if I’m using DKI, just to keep the misspelled words out of ad copy. Thanks for your comment!

  4. I’ve moved to using quite a few broad match modified keywords rather than try to go nuts with all the possible variations of exact and phrase. That way I don’t miss much and can extract any gems that need to be isolated and targeted.

    I am so glad that Melissa said she has some agroups with a large number of keywords. I agree that if they are close enough that should be ok.

    I couldn’t tell if most people approve or disapprove of that monster account. It’s far larger than anything I have worked on. But I have to admit that over a period of a couple of years I compiled a list of more than 1,000 misspellings for a single very competitive ($100+ CPC) keyword.

    If the monster account is using negs to avoid bid overlap, is that really worth the effort? I suppose it could be if it prevents something less target from showing ads instead of the same thing in an exact match.

    After reading the comments my thought was, “Well, what are the STATS for that account?” Is someone wasting effort to be anal about things or is that one rocking account?

  5. Hi Mellissa, the way we do it (almost as Mark suggested) is by figuring out search intent of every keyword and only put keywords together that share the same intent. The idea behind being that if people are searching for the same thing, they can be served the same ad, and thus the ad group is formed. Then, there’s no need to worry about the size of the ad group since all the keywords share the same search intent. We solved this in our tool (which I won’t pitch here) to do it automatically in a single click for as many keywords as we want. I can’t remember the last time I had to manually group keywords.

  6. Granularity is key for me so I have one keyword per adgroup where possible. It might appear to take a lot of time but it doesn’t. It means that I have the keyword in the headline, description and/or display url if i want to. You can also manage your cross match negatives better. I tend to find better efficiency with this approach.

    • Melissa Mackey says

      Really? That’s interesting. Have you done this with a large account (>10,000 keywords) and still found it to be more efficient? I’m curious. Thanks for your comment!

      • That is the approach we take where feasible. There are some adgroups with about 5 keywords but that is the max i’m prepared to take. if you use a bidding tool like doubleclick, optimisation becomes a lot more easier. One easy way of achieve this level of adgrouping is to categorise your keywords to a point and then append the keyword to the grouping to achieve one keyword per ad group. You can then manage the difference in performance very easily. SQR, cross match negatives etc becomes easy. I only use exact and broad match modifier. I don’t mix match type in the same campaign either.

Speak Your Mind

*

This site uses Akismet to reduce spam. Learn how your comment data is processed.