PPC Chat Streamcap – Google AdWords Keyword Tool

Hello again!

Our host Matt Umbro (@Matt_Umbro) put together a great set of questions on this week’s pay-per-click topic on the “Google AdWords Keyword Tool”. The following is the transcribed Streamcap from the live chat.

Q1: Do you find value in the AdWords Keyword Tool? Why or why not?

  • Yeah, its not the worst keyword tool out there. – John Lavin (@johnnyjetfan)
  • Value depending on how niche the keyword/client. Very niche = less value. – Andy Groller (@AndyGroller)
  • Definitely valuable. Gives some keywords, but mostly starts branches of keyword to look into. Great for negatives too. – Michelle Morgan (@michellemsem)
  • It provides some value. Very high level data I use when pitching new business. – Justin Freid (@Justin_Freid)
  • Value? Sure…it’s a great starting point for your paid campaigns! Now, accuracy…that’s a whole different ballgame. – Cleofe Betancourt (@askppc)
  • Yes, to an extent. I find more value in the LP & Client feedback, but GKWT is a good indicator of high level concepts. – Chris Kostecki (@chriskos)
  • Great tool for finding some basic ideas, estimates & finding negatives! Great for (rough) estimates, but not an end-all. – Aaron Levy (@bigalittlea)
    • I agree, when searching for keywords it’ll make suggestions for word I don’t want so I’ll add them as negatives. – Harris Neifield (@HarrisNeifield)
  • The keyword tool is OK and good for a baseline, but like any other keyword tool take the data with a grain of salt. – Matt Umbro (@Matt_Umbro)
  • Yes, the keyword suggestions help with expansions and also give some colloquial insights for British & Canadian advertising. – Harris Neifield
  • To an extent – the tool is good for broad term research, not so much for niche terms. Good for negative keyword research though. – Jessica Cameron Ruud (@camruud)
  • Still my go to but discrepancies between signed in and signed out values still a concern. – SEMantiks (@SEMantiks)
  • Needless to say, the estimated traffic columns go completely ignored. – Michelle Morgan ++++
    • Yep – the numbers are crazy inaccurate. – Jessica Cameron Ruud
  • Grain of salt especially since it only shows data for keywords that have ads displaying. – Andy Groller
  • Personally I take it as a rough estimate, especially where CPC is concerned. I apply 60-70% of est to be more accurate. – James Luty (@jamesluty)
  • Very unstable and somewhat uncomfortable for big loads of work, but I haven’t anything better (even paid solutions). – Sergey Smirnov (@Smirnovi4)
  • It is helpful for adding keywords but sometimes more irrelevant keywords are given than relevant (good 4 neg keywords). It can cause headaches. – PurePPCcom (@pureppccom)
  • Yes + Wordstream and Marin Search Marketer. – Mike Shollenberger (@webjock)
  • I find value in the directions it at times points me toward but generally not in the actual keywords it produces. – Amy Hoffman (@Hoffman8) +
    • Same here. Mostly just gives ideas on other places to expand. – Michelle Morgan
  • I don’t use the tool to make estimates, but it’s definitely a standard part of keyword research. – Roger Sikes (@rogersikes)
  • It also helps to see what Google deems similar. Sometimes the keywords are waaaaay off. – Amy Hoffman ++
    • True enough…its more “miss” than “hit” most times. – Cleofe Betancourt
  • Use it as a guide for budget proportions & campaign separation, but not a “WE WILL SPEND THIS” estimate. – Aaron Levy
  • Sometimes also a good initial part of research to show clients. Just to manage their expectations, and direction. – James Luty
  • What drives me nuts, is for existing accounts, it will suggest keywords I already have as negatives. Needs a filter. – Michelle Morgan +++
  • I use GKWT more for negatives than keywords to add. And ideas for ad group themes, now that Wonder Wheel is gone. – Lisa Sanner (@LisaSanner)
  • One thing that annoys me is the lack of attribution to quality score. it makes CPC’s even more redundant! ALSO! it isn’t clear how the volume is shared…i.e. G search / partners or both. I would assume estimates are for all networks. – James Luty ++
    • Great point! Again, more information we’ll probably never get to know. – Michelle Morgan

Q1 Summary: It’s probably safe to say that most/all PPCChatters feel that the Google AdWords Keyword Tool is valuable to an extent, but you can’t trust the data 100%. It tends to be good for finding Middle and Head terms, but is severely inadequate for identifying Long-tail keywords.

The GAKT is also a good source for identifying negative keywords that will make up the bulk of your unwanted clicks and impressions. You will still need to do other negative keyword research like run Search Query Reports regularly.

*Note* It is also important to understand that the Google Keyword Tool statistics are for AdWords advertisers. This data is probably skewed based on that fact. Many search marketing professionals assume that the numbers and keyword variations provided might be based off of all Google Network Search Queries, even those that do not contain PPC ads. This is probably not the case.

Q2: Do you share Keyword Tool data with clients, especially when they are curious about average cost-per-clicks (CPCs)?

  • Yes, but it’s always included with about 90 caveats. Its more to explain proportions than anything. – Aaron Levy
  • We’ll share CPC ranges but caution multiple times on it being ballpark; actual CPCs go way beyond info in tool. – Andy Groller
  • No, Clients are on a need-to-know basis. They don’t need to know about the tools I use, I think it would just confuse them. – John Lavin
  • Accuracy for $ & Traffic tends to be way off and I’m very frank with clients about that. – Amy Hoffman ++
    • Anybody give estimates on % off? – Ryan Campbell (@_ryancampbell)
      • Never tried to estimate it really. Just used trial and error and usually forgot the original suggestion. – Michelle Morgan
  • I multiply results by .1 and get more or less real picture for most niches – only then show to customers. – Sergey Smirnov
  • Yes, I use it in proposal, but it is made clear they are estimates. But anything after that I do not share from Keyword tool. – Justin Freid
  • Still it does lend itself to which keywords will have more volume – even if the estimated volume is wrong (which it usually is). – Amy Hoffman
  • I don’t like to share estimated searches and costs, just raises more questions and unnecessary worry. – Matt Umbro +
  • No, I don’t share it. If its a pushed topic I’ll mention it, but also hammer home how inaccurate the tool can be. – Michelle Morgan
  • No, I may use it as a signal to ballpark the costs – not share directly, but it is too far off from what I see in live accounts. – Chris Kostecki +
  • Not if I can help it. It prompts too many unanswerable questions. – Lisa Sanner
  • Yes, for individual words CPC’s can be off but with a few disclaimers data is shared. – Harris Neifield
  • Accuracy, just like impression estimates, is not very accurate for CPC. – Justin Freid
  • We share CPC data sometimes from the keyword tool but not unless we are sure it’s in the right ballpark. Sometimes it’s way off. – pureppccom
  • We are careful about the CPC tool especially because it sets expectations incorrectly. I tell them about previous experience. – Amy Hoffman
  • I trust my own experience and CPC estimations over the keyword tool. But it’s taken 2+ years to get there. – Neil Sorenson (@iNeils)
  • Sharing est. CPC can put you in good stead in over delivering – but it must be clear that these are estimates. Nothing more. – James Luty
  • The day I rely on Google for my account insight (besides the raw data) is the day I’ll have a fork stuck in my eye. – Chris Kostecki ++++
    • I wonder why Google can’t make these numbers more accurate. I mean, they have the data. – Matt Umbro
      • They like things to be as opaque as possible. They can charge us for what we don’t know. – Michelle Morgan
  • No, b/c then they want to bid on keywords that bring unqualified traffic, are too broad & incur too high of costs 4 budget. – Mike McGinley (@Mike_McGinley)
  • The CPC estimates have never delivered for me. Nearly every time I’ve tried to abide, I’ve been below 1st page bid. How close or far the tool’s suggestions (especially traffic) also heavily depend on geotargeting. – Amy Hoffman
    • Geo-estimates = useless. – Aaron Levy
    • I’ve seen this too. – Harris Neifield
  • Does anyone know if Google has an assumed QS when giving bid estimates? – Neil Sorenson
    • That would be very interesting to know. Would make a huge difference. – Michelle Morgan
    • Pretty sure its an overall comp. average. – Aaron Levy
  • Initially yes to get a VERY general idea for possible opportunities but it’s clearly noted they are highly loose estimates. – Cassandra McClure (@imcassy)

Q2 Summary: In general, this tool is too inaccurate to base campaign pricing, volume, and expected results on. Therefore most PPC professionals try to stay away from providing this data to clients because it opens up an entire additional conversation about accuracy and expectations. It also makes up look less than experts when we are trying our best to provide solid advice to clients and data from our main source is statistically unreliable:(

Q3: What is your first impression when the Keyword Tool shows a huge drop-off from a broad to phrase match term? Ex: if a broad match keyword shows 10,000 monthly searches, but then drops down to 100 phrase match monthly searches?

  • That Google realllllyyy wants you to use broad match and spend lots of money! – Justin Freid
  • The broad term likely has multiple meanings depending on word arrangement – reconsider your keyword selection! – Aaron Levy
  • There are a lot of negative phrases that need to be added or huge opportunities for long tails or a whole lot of misspellings and variations (i.e. apartments, apts, apt rentals, etc). – Andy Groller
  • I generally think that the broad term for that keyword is canvassing an audience that is too broad. – Amy Hoffman
  • I just dismiss it as Google trying to hoodwink me into using only broad. They can forget it. – Michelle Morgan
  • It shows availability of junk queries: like Free Smth that would account for 80-90% of the broad searches also queries like how to, where to, what is or crossings with other meaning. – Sergey Smirnov
  • The fact that lots of negatives need be added before activating is a good point also. – Amy Hoffman
  • Broad Match is the simplest match to understand for newbies. Maybe Google is just trying to make it easier for newbies to get? – John Lavin
  • Remember this depends on how many terms are put in the tool at once. If a long list of tail terms is used phrase match will drop. – Harris Neifield
  • I may use the broad keyword w/ a very low bid and really amp up negative keyword list. – Mike McGinley
  • It’s an indication to start with a modified broad match and baby sit search query reporting. – Lisa Sanner +
  • When there’s a big drop-off from broad to phrase I think you have to look at the broad term & see if it’s a natural looking keyword. Often times unnatural looking keywords are getting broad searches from different terms being paired w/ broad match. – Neil Sorenson
  • Entirely depends on the term I think. & how you structure your campaigns. I would use exact and phrase. targeted and lower cpc. – James Luty
  • If broad is that large and exact is that small it tells me that most of what broad is matching on is not very relevant to it. – Kiko Correa (@obiwankikobi)

Aaron Levy’s post is relevant here:

Does Google Promote What They Preach?

Q3 Summary: Because of the way that AdWords matches many other search queries back to an advertisers Extended Broad Match keywords, and the additional terms that will be matched with Phrase Match keywords, we need to realize that there are going to be huge differences in search volume numbers between broad, phrase and exact keywords. The difference in volume data between match type is often more profound for 1 and 2 word keywords than for those with 3 and 4 words.

Q4: Why doesn’t Google show Modified Broad Match Keyword stats (or at least indicate the search volume and costs would be different)?

  • They don’t fully get it yet either? – John Lavin +
  • It’s too new! It took them years to update estimating features, lord knows how a new match type will work. – Aaron Levy
  • My opinion – the Modified broad match is simply to young to get into the keyword tool. – Sergey Smirnov
  • Money, money, money, money…. money. Is that a sufficient answer? – Andy Groller +++
  • Perhaps not to confuse advertisers further, but I don’t necessarily think Google would find this a problem. – Matt Umbro
  • If Google can’t even give accurate traffic and CPC estimates for regular keywords on broad match, they’ll screw it up even worse for MBM. – Neil Sorenson
  • They want us to add that later – on top of broad – not instead of broad and save money. – Amy Hoffman
  • Speculation – less ad revenue and because many advertisers don’t understand what modified broad match is. – Harris Neifield
    • Many small business owners don’t even know about it, let alone understand it. – Justin Freid
  • I want to say they don’t have enough data. But the data they have for established match types is crap. So I’m at a loss. – Michelle Morgan
  • MBM is like the broad match of 5 yrs ago. They couldn’t get estimates right then either. Only give us more to complain about. – Lisa Sanner
  • They want attractive volumes to show any advertiser. Agreed they also want the money – they are like any other profit org. – James Luty
  • Google’s interest are to drive costs/activity, they’ve something to lose if its too accurate (& if not accurate enough). – Chris Kostecki
  • G blame the rise of CPC’s on agencies – I’d say its more from misguided info from them! Bid wars but also focus on QS. – James Luty

Q4 Summary: Modified Broad Match has many different matching options that makes it difficult to predict keyword data. With BMM you can choose to Anchor 1, multiple, or all of the words in the targeted keyword For instance, the keyword “Pay Per Click” could be BMM’ed 9 different ways just in this word order: +Pay +Per +Click, +Pay Per Click, Pay +Per Click, Pay Per +Click, +Pay +Per Click, +Pay Per +Click, Pay +Per +Click.

You can also change the word order around 5 other ways and then modify these 9 different ways: Pay Click Per, Click Pay Per, Click Per Pay, Per Click Pay, and Per Pay Click.

Google would have a difficult time estimating traffic and CPCs for all the possible variations. Unless they just supplied data based on the fully modified version, exactly as it was typed in the box: Pay Per Click = +Pay +Per +Click. 🙂

Q5: Discuss a unique way you use the Keyword Tool that others might not know?

  • Blend the keyword tool findings with Insights to get an estimate of seasonality & future volume. – Aaron Levy
  • I wouldn’t say others don’t know about it, but if I’m using it, I’m mainly looking at competition & possible niche opportunities. – Cassandra McClure
  • Others have hit on it, but good for finding negative keywords. – Matt Umbro
  • Use the keyword tool to get ideas about niches we may not have thought of and to kind of gauge how people are searching for our brand and our competitors. – Amy Hoffman +
  • I use the website section & use informational sites (Wiki, About.com) to find keywords people may have seen during research. – Michelle Morgan
  • Exploring other language markets in the US, and coupling it with the traffic estimator for loose insight of new opportunities. – Chris Kostecki
  • CTD – use the tool to find new niches, dump them into soovle.com to find even more! – Aaron Levy
  • Use URL to estimate search terms. Manually then modify this for a more precise research session. Great for finding long tail. – James Luty
  • Put broad single words in and can come up with some ad group themes that are more peripheral. Then research those separately. – Lisa Sanner
  • Adjust geo-targeting to get regional difference info rather than using Google insights, which is all SEO+PPC. – Harris Neifield
  • Use the keyword tool in combination with Google suggest and ‘related search terms’ at the bottom of the search page. – Nihkil Inamdar (@NikhilInamdar)
  • Though it’s not all you should use, I use the URL feature to help predict what the QS situation will be. Not enough chars! It can help you fine tune your page content for relevance (and hopefully higher QS). – Roger Sikes
    • Not a huge fan of the URL feature in general…it seems to return data on everything except the site content. – Cleofe Betancourt

Q5.1 Does anyone check stats comparing desktop/laptop searches to tablets/smart phones?

  • A bit….just starting to target these devices in adwords. – Cleofe Betancourt
  • Could do that when building tablet campaigns, but I like to separate tablets and desktops. – Harris Neifield

Q5 Summary: There are too many good tips here to do a summary so I’ll add one of my own instead…

If you have the “Local Search Trends” column activated during keyword research, you can download a keyword spreadsheet that will contain the monthly search volume for the previous 12 months. This will give you high and low estimates in numerical form.

Q6: Do you feel the Keyword Tool has made enough improvements over the years to warrant more trust from advertisers?

  • Barely. – Melissa Mackey (@Mel66)
  • No. – Harris Neifield
  • Haven’t used it for “years”. The time that I have used it, I only know I can’t trust it. – Michelle Morgan
  • In a word, nope! It’s inherently flawed bc it pulls HISTORICAL data across the board. I’ll never trust it. – Aaron Levy
  • Nope – Cassandra McClure
  • Small improvements, but not much more trusted gained. – Joe Kerschbaum (@JoeKerschbaum)
  • No – they’ve added things but no one seems to find their estimates accurate so.. no love from me. – Amy Hoffman
  • Not at all. If anything, with G search changes, the keyword tool has fallen even more behind! Trust has fallen for me, with poor support. – James Luty
  • Improvements have been made but until most of the big problems are resolved, my trust will always be very limited. – pureppccom
  • It HAS improved over time. C’mon, think back to 2007 or 2008. – Melissa Mackey
  • They’ve added a lot of refinement, but need to get more accurate at the granular level, the data is flawed so the tool is flawed. – Chris Kostecki
  • A resounding no. tols dont equal trust. – John Lavin
  • Steps in right direction but long way to go before more trust put into it. – Andy Groller +
  • Ehh, moving from an F to a D average doesn’t make someone a good student to rely on. – Harris Neifield ++
  • This is old skool: but the Overture keyword tool was pretty awesome. Interface wasn’t great, but the estimates were sorta close. – Joe Kerschbaum

Q6 Summary: Most PPCChatters still don’t trust it very much. However, I do feel that, starting with it’s big update late last year, that Google has made some progress and I do trust it’s data more in 2011 than in any year past. Especially the “Approximate CPC” estimates, which use to be a complete joke. Now they are kind of in the same ballpark.

Q7: What data or feature would you like the Keyword Tool to show that it currently does not?

  • Dedupe from my current keyword list!! – Melissa Mackey +++
  • As discussed earlier modified broad match estimates. – Harris Neifield
  • Look at my current negatives and take them into account. – Justin Freid ++
  • Accuracy! That’s it, put int he keywords and tell me how much traffic I can expect, and how much competition there is. – Chris Kostecki
  • Search partners vs Google search, modified broad. Accuracy more than anything though. – Andy Groller ++
  • Ranges for CPC estimates. Where’d my high and low go?!?! – Aaron Levy ++
  • I wish exporting data from multiple match types was cleaner. – Matt Umbro
  • Stemmed & vertical keyword lists; correct traffic estimates; negative keywords; combo keyword suggestions. – Joe Kerschbaum
  • Average CTR, CTR split between Organic and PPC. – Sergey Smirnov
  • I would love the keyword tool to somehow incorporate something that would show what your quality score would be on certain keywords. – pureppccom
  • Typical conversion rate would be helpful but hard to do since everyone has diff conversion types. Display network info would be nice, too. Andy Groller gave me that idea b/c search partners vs Google would be lovely! – Amy Hoffman
  • What would be harder to get – accurate AdWords estimates or a balanced federal budget? – Robert Brady (@Robert_Brady)
    • Any other miracle requests for today? – Sergey Smirnov

Q7 Summary: There are some very good suggestions here. Deduplication from a current list and taking into account current negative keywords would both save advertisers some serious time.

I would also have to echo those that are looking for better Cost Per Click estimates. For Approximate CPC estimates, Google AdWords currently states that “This is the approximate cost-per-click you might pay if you were to bid on the keyword. The CPC is averaged over all the ad positions.” In my opinion, this is too broad of a range as it also includes the CPC for clicks from positions 9+. I personally would like to see this be a more accurate average of the top 5 or top 8, where almost all of the traffic is generated from. This would provide advertisers more realistic tragic costs.

Q8: When beginning a campaign, does the Keyword tool help you determine what match types to put your keywords in?

  • Nope – Joe Kerschbaum
  • Nope. Just ideas for negatives and places to look for further research. – Michelle Morgan
  • Absolutely – it gives a good idea of how far afield Google’s matching might go. – Melissa Mackey
  • Nah, we decide that based upon the client’s needs and strategy. – Amy Hoffman
  • No, it wants you to use broad match. I usually go on hunch/experience and understanding the clients needs/business. – Justin Freid
  • Not always. We use the data the tool gives us and decide what we want to do with it on our own – based on our experience. – pureppccom
  • Not for me. I use all three match types in the beginning and dwindle it down from there. Keyword tool only gets me some ideas and keywords. – John Lavin
  • More about negatives than match type; common sense typically prevails. – Andy Groller
  • Not really. – Robert Brady
  • Sorta? At least it makes us think twice about the super high volume terms. – Aaron Levy ++
  • Not at all. – Sergey Smirnov
  • The tool can help to determine if a term will be in broad match, but other than that strategy and commen sense prevail. – Matt Umbro
  • How many of you will use mostly all 3 match types to begin with? (situational based of course) – Neil Sorenson
    • Not often, usually expand from phrase. – Chris Kostecki
    • {Hand raised} – Andy Groller
    • I usually start with MBM unless theres a reason not to. – Aaron Levy
      • True, MBM typically used more than broad to start. – Andy Groller
    • I would – Michelle Morgan
    • I find myself going back and forth on whether or not to use all 3 match types to begin w/ but like the idea of starting only w/ MBM. – Neil Sorenson

Q8 Summary: Far and wide the answer is, and should be, NO. The data is just not reliable enough to be actionable for keyword matching, and once you take into account the Quality Scores that your new keywords will have, #FoGettaBoutIt.

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.

Participants

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

About the Author

James Svoboda provides PPC Consulting at WebRankingThis is a guest post by James Svoboda, managing partner at WebRanking, Sphinn Editor, infrequent search marketing blogger, SEM content hound, Google+ heretic, and Co-Founder of the Minnesota Search Engine Marketing Association, and tweeter @Realicity.

Be Sociable, Share!

Tags: , , , , ,

4 Responses to PPC Chat Streamcap – Google AdWords Keyword Tool

  1. […] The best example is the #ppcchat hashtag conversation that happens every Tuesday or so. Here’s a streamcap from a recent chat. This is a good list of PPC people on Twitter to start […]

  2. […] The best example is the #ppcchat hashtag conversation that happens every Tuesday or so. Here’s a streamcap from a recent chat. This is a good list of PPC people on Twitter to start […]

  3. […] The best example is the #ppcchat hashtag conversation that happens every Tuesday or so. Here’s a streamcap from a recent chat. This is a good list of PPC people on Twitter to start […]

  4. […] The best example is the #ppcchat hashtag conversation that happens every Tuesday or so. Here’s a streamcap from a recent chat. This is a good list of PPC people on Twitter to start […]

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