PPC Chat Streamcap – Keyword Matching & Control

This week James Svoboda (@Realicity) came up with yet another great question set titled “Keyword Matching & Control.” The following is the transcribed Streamcap from the live chat:

Q1: Now that @adCenter has BMM, what Percentage of Broad, BMM, Phrase and Exact Match Keywords typically make up your Campaigns?

  • I would estimate that I typically use 55% Exact Match, 40% BMM, 5% Phrase and 0% Broad. Yep, Zero normal broad match keywords. – James Svoboda (@Realicity)
  • Evenly split across BMM, phrase, & exact, with a few broad thrown in on brand terms. – Melissa Mackey (@Mel66)
  • We typically have an even number of all, except BMM now takes over most (if not all) broad match keywords. – PurePPC.com (@pureppccom)
  • Totally depends. Some of my accounts run 1/4 each. Some are heavier on exact and phrase. New accounts are heavier on BMM. – Heather Cooan (@HeatherCooan)
  • largely depends on the account – if it’s growth, edges more towards BMM. control, P/E – Aaron Levy (@bigalittlea)
  • SQs on broad match brand terms can be enlightening & alert to potential issues as well as drive leads. – Melissa Mackey
  • 60% exact 40% BMM. Always end up having more exacts that any other match type. – Jordan Glasner (@glasner)
  • It depends on the goals, results, budget of each client, so case by case. But broad is very low and BMM is higher now. – Mark Kennedy (@markkennedysem)
  • Usually start off with 25% each (and a healthy negative kw list), then narrow down later. – Michelle Morgan (@michellemsem)
  • I have found broad match to be useful in digging up competitors I didn’t realize were there. But watch close! – Heather Cooan
  • Q1.1 Follow-up: For those using more Phrase match than me, how does your click/impr split come in vs Exact & BMM? – James Svoboda
    • I’ve seen phrase imp go way down when also using Exact & BMM. – Melissa Mackey
      • Same here. That’s why I use it less. – James Svoboda 
    • Depends what kind of historical KW data I have to start w/. I’ll use BMM mainly as discovery/reach w/ converting Exacts – Andrew Baker (@AndrewBaker72)
  • It’s probably 25% BMM, 74% Exact, and 1% Phrase. I think I have 1 or 2 true broad match keywords. – Dennis Petretti (@Denetti)

Q2: Now that @adwords has Near Phrase and Near Exact keyword options, do you intend on using them and why?

  • Yes but on a VERY limited basis until I see how they perform. – Melissa Mackey
  • Yes where word order matters & want to expand. But mostly, no. – Aaron Levy
  • I honestly don’t see a need with all of the other match types available. – Heather Cooan
  • Turning off for most, leaving on temporarily for select few based on search volume need. SQ reports will be closely mined. – Mark Kennedy
  • Hesitantly will try and use them because there is so much KW variation in our space and we want want MORE volume. – Neil Sorenson (@iNeils)
  • Yes – through much testing I imagine I will use them. But NOT by default. – Rick Galan (@RickGalan)
  • No. I would if they were a modifier like BMM, but not for a campaign wide setting. You lose too much Exact control. – James Svoboda
  • Broadly, I use whatever match type scores the highest Qual. Score. I found the kywds 4 our campiagns that perform the best before all this Match Types discussion got big – Gnosis Media Group (@GnosisArts)
  • No, I don’t intend on using them. I don’t use phrase much, and I don’t like near exact. Exact should be just that, exact. – Dennis Petretti
  • I’m opting out on all our accounts and setting up several tests to see the results before deciding the best account set-up. – Andrew Baker
  • No I immediately turned it off in all my accounts. I want full control, not Google. – Sebastiaan (@SebastiaanHu)

Q3: What Percentage of 1 & 2, 3 & 4 and 5+ word keywords typically make up your Campaigns?

  • I use maybe 85% 3 & 4 word keywords, 10-15% 5+ word keywords, and less than 5% of 1 & 2 word keywords. – James Svoboda
  • The bulk of my keywords usually lie within the 3 & 4 word length. 1 & 2 – very few, 5+ somewhere inbetween. – PurePPC.com
  • I’d say most fall into 3-4, but I’d have to really check to get the actual proportions. – Mark Kennedy
  • Good Q, I’ve been messing around w/ a custom report in Analytics to pull this data – not working yet so difficult to say. – Andrew Baker
  • Sorry I’m late, gr8 question…while a higher percentage of my KWs are LT, their impressions are shrinking. – Chris Kostecki (@chriskos)
  • I wouldn’t even know – I just use the keywords that convert. Never bothered counting percentages. But if I had to guess, I think the maj would be 3-4 word keywords. Actually, as I look at a campaign now, there were quite a few 2 word PPC conversions – Gnosis
  • In older accounts I’d say 80%+ 4-5 and 10% 5+…I run a lot of longtail. – Heather Cooan
  • 1: <1% 2: Lots 3: Even more 4: Some 5: few. – Aaron Levy
  • 1: 0%, 2: 5%, 3: 60%, 4: 30%, 5+: 5% – Jordan Glasner
  • Percentage wise similar to you, but spend/traffic/revenue wise a lot more on the 2-3 word keywords. – Richard Fergie (@RichardFergie)
  • It definitely varies by industry. I’ve seen strange trends of only [2-3] word queries vs mostly [4-5] word queries. – Cassie Allinger (@CassieAllinger)
  • Depends how you are targeting many converting KWs could be 6/7/8/9 words – only way to target them is BMM, Exact/Phrase = LSV. – Andrew Baker
  • Median is 3. Average is probably like 3.5 – Robert Brady (@robert_brady)

Q4: How do you use and mix Broad, Phrase & Exact Match Negative Keywords for your Campaigns?

  • I use all three match types for negs based on the SQ report. Some campaign, some adgroup level. Mostly broad, though. – Mark Kennedy
  • Usually anything that’s two words goes phrase and anything one word goes broad. I do use a mix when trying to protect adgroups and campaigns from each other though. – Heather Cooan
  • I use all 3 match types to box in & box out, esp when I have separate ad groups by match type. – Melissa Mackey
  • I’d rather see stricter positive matching, as opposed to relying heavily on negatives to control loose matches. – Chris Kostecki
  • Exact and BMM in separate ad groups. Then use negative embedded match to stop BMM ad group from hitting exact match. – Jordan Glasner
  • I mostly use Broad with some Phrase for negativing out select queries, and rarely use Exact negatives except for shifting Imps. – James Svoboda
  • The majority of the negative keywords, at any level not just campaign, are broad match negatives. – Dennis Petretti
  • In generally I use exacted match negatives. But for the more general negatives I use phrase. I don’t use broad. – Sebastiaan
  • Broad for growth, MBM for cautious growth, Phrase for controlled expansion, Exact for tight control. Mix based on priorities! – Aaron Levy
    • I always looked at it: BM: Blind Discovery, MBM: Discovery, Phrase: Opportunity, Exact: Qualified. – Chris Kostecki
  • A lot of Exact Match Negs at AG level when using match type AGs. – Andrew Baker
  • Awhile back I switched to mostly phrase match negative keywords and feel good about it. Still use exact neg KWs, no broad. – Neil Sorenson

Q5: How do you address Geo-Specific Keywords for Non-Geo-Specific campaigns?

  • Give them their own camp/adg with custom ads and landing pages. – Michelle Morgan
  • Tackling this problem with a client soon, may test location insertion. – Aaron Levy
  • For Ecommerce campaigns geo-keywords can tank conversion rates a searchers tend to be looking for local stores. – James Svoboda
  • They go into separate adgroups or campaigns and I cross reference negs for control and include the modifier in ad copy. – Heather Cooan
  • For Non-Ecommerce campaigns, geo-keywords often have less of a negative impact on conversions and can be acceptable to use. – James Svoboda
  • Problems I usually run into A: low search volume, B: too many geo’s to name C: people like to search by zip, not city. – Aaron Levy
    • Interesting. I don’t often see zips showing in search queries… maybe Goog is changing those to cities? – James Svoboda
      • Mostly for hyper specific businesses. Family planning/house cleaning/exterminators/catering/dating. stuff like that! – Aaron Levy
  • Also having issues with names that have multiple locations. Oakland NJ vs Oakland CA. Google always assumes you mean the big one. – Mark Kennedy
    • Shut off the "search intent" targeting tool, should solve the problem! – Aaron Levy
      • Did that, tried all 3 settings. Still an issue. Set Google to NJ and search "oakland dui laywer". will see CA ads. The workaround was to just use a NJ & Kw ad, but can’t use oaklyn dui lawyer as KW. Still trying other things, though. – Mark Kennedy
        • Yeahhh I kind of see what you mean. Maybe try MBM-ing it and leave the state abbrev unmodified? – Aaron Levy
    • Start excluding those locations in the location targeting section. – James Svoboda
  • Still working on this, geo-specific keywords and negative keywords is going to be one of my summer projects this year. – Dennis Petretti
  • For Geo, try to find terms that are relevant to area and have scale (ex., boston for eastern mass, berkshire for western mass). Searchers understand they have to broaden their area for relevant results (we are all victims to Google’s macro approach) – Chris Kostecki
  • Having decent luck w/large city names as geo modifiers. – Melissa Mackey

Q6: Beyond Country & State targeting, how do you use Location Targeting & Settings for PPC control? 

  • If targeting a specific metro (unique message/event/offer), will negative it out of broader campaigns. – Chris Kostecki
  • I’d rather start big and exclude (i.e. target a metro and exclude cities) rather than pick and chose. More volume that way! – Aaron Levy
    • True, but can add work for geo-keyword campaigns like philly dentist when trying to target the burbs also. – James Svoboda
  • Use city/metro all the time for local clients. – Melissa Mackey
  • Reverse answer: I *don’t* use any setting that determines "search intent". I’ll do that. G has a hard enough time knowing location. – Michelle Morgan
  • Now that the new zip code feature is back, I’ve been using that for the SMBs. No qualms so far, but still not 100% accurate. – Mark Kennedy
  • I find that using Negative Locations for cities & states can greatly improve control. Example: Portland Oregon & Portland Maine. – James Svoboda
  • Also use city/zip targeting to drive foot traffic w/specific message, diff message for broader geos and click to call campaigns work great in local areas when they have a local number. – Melissa Mackey
  • For us the main geo-targeting method beyond state is by USDA growing zone. I wish AdWords had a setting for that. – Dennis Petretti
  • I like to get granular for local specific ad copy, but struggle with volumes on search. Wrks great for display! – Heather Cooan
  • Target for weather, which has to be broken down smaller than state. Especially "good" for major storms. – Jordan Glasner
  • Bid by UPS zone for heavy items that require huge shipping costs. – Jordan Glasner
  • If you’re in ecomm and ship to "US only", don’t shy away from military bases overseas. USPS ships there and conv rate is HUGE. – Aaron Levy

Q7: How would you change/fix/blow-up Keyword Match Types to address Control and Volume?

  • No more smarmy default settings to "improve" our accounts. – Aaron Levy
  • I would rebuild Phrase Match as it is useless for longer 4+ word keywords as you need the words to be in an exact in order. – James Svoboda
  • A keyword without any punctuation should be exact match. Good for new advertisers, *very* bad for Google. – Jordan Glasner
  • I’d love a phrase plus BMM combo keyword. Something like +blue "tennis shoes" – Dennis Petretti
  • I’d maked Modified Phrase -> "keyword keyword * keyword" , where the * can be any word. – Paul Kragthorpe (@PaulKragthorpe)
  • Keep KW Matches at the KW Level! – Chris Kostecki

Q8: How does Device Targeting (desktop, tablet, mobile) affect your keyword matching and control?

  • I tend not to worry as much with Mobile since ppl basically only search what suggest… suggests. – Aaron Levy
  • I’ll try shorter 2 + 3 word keywords for mobile campaigns that I wouldn’t normally for desktop. – James Svoboda
  • Separate campaigns for every device type, but use same march type strategy. Keyword lengths for mobile are 1-2 words shorter. – Jordan Glasner
  • For large volume targets I separate (& see different bidding patterns), also Tablet is truely a 3rd device and not a mobile variant. – Chris Kostecki


More PPCChats

Don’t forget to stay tuned for the next #PPCchat on Tuesday at 12 noon Eastern, 9 am Pacific and 5pm in the UK. Same Chat time, same Chat channel.

CRO Chat

Join us on Thursdays at 12 noon Eastern, 9am Pacific and 5pm in the UK for #CROchat where we talk all about Conversion Rate Optimization.


Check out the PPCChat Twitter list to see and connect with all current and prior participants.

About the Streamcaptain of PPC

This is a guest post by Paul Kragthorpe; Search Manager for WebRanking, SEO in Minneapolis, MN, #PPCChat Streamcap Grabber, Search Marketing Blog Writer. Connect with me @PaulKragthorpe, and Google Plus.


Be Sociable, Share!

Tags: , , , , , , ,

Leave a Reply

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

Recent Tweets

I am speaking at SMX East
PPCChat.co was rated one of the Best PPC Blogs by Boost CTR