• Neptune Moon


    Thanks for sharing this.

    Here is my question for the lead gen accounts – what will be the impact on QUALITY of leads/conversions? Anyone who works in lead gen knows that the quality of the conversion is so important. I’d take a lot fewer conversions if the quality is better every day of the week. I don’t want more conversions that are ultimately crappy leads.

    In a similar vein, it will be interesting to see how lifetime value of e-commerce customers compare too and if those stats go down as close variant conversions increase (if in fact they do even increase for all accounts).

    Google seems to lack the vision or understanding that what happens after the click is actually more important to businesses than what happens to create the click. If acquisition costs increase and conversion quality goes down, advertisers will not be satisfied with just having “more volume” or “more conversions”. You’d think they’d get that a little more.

    So interested to see how all this plays out and to see data from others, especially those like Wordstream who have access to such massive amounts of account data!

    • Sam Owen

      Hi Julie,

      Thanks for the comment. I agree and eluded to it slightly in my post (although didn’t build it out) – are leads from people who can’t spell as valuable as from people who can spell? Maybe, but then again there’s a good chance they aren’t. I don’t have the data to say either way unfortunately (CRM tracks keywords not queries).

      I understand Google’s vision to keep everything simple but I don’t understand why simplicity needs to come at the expense of functionality. There are ways to keep things straightforward without taking away from the tools power users have.

  • Susan Wenograd

    Hi Sam!

    Curious: was the negative keywording (yes, I just made that a verb) more aggressive in the accounts where the exact match variants did ok? Or would you say it’s about the same across the board?

    Just wondering if that had any impact on the back end before the queries even hit.

    • Sam Owen

      I can tell you as a standard we don’t negative out misspellings or plurals generally (yet).

      However, some accounts had more plurals built out in the first place. My guess is that the fewer plurals built out, the larger the impact of CVs.

      Thanks for commenting!

      • Susan Wenograd

        Gotcha. I’m not totally melting down over this. It’s really annoying, but ultimately it just means I’ll have to spend even more time negative keywording every day. Sigh.

        • Sam Owen

          I think that’s the sensible approach. As almost all Hanapin clients have CVs turned on, I don’t see a major drop in performance being an issue. I do feel for people who’ve analyzed their data and decided they need to be turned off though.

  • Melissa Mackey

    Just ran a quick analysis on a very large lead gen client. For exact match close variants, we saw an 11% higher CPA. For phrase match close variants, we saw a whopping 35% higher CPA. We are also seeing a higher CPC for the close variants than for the matched keywords. But, we got 20% more conversions from using variants at only a 5% increase in CPA. So on the surface it seems worth it.
    However, several of the client’s campaigns are budget limited. Seems to me we could shut off the variants and get more conversions without sacrificing CPA. This would actually be a good test – and an option I won’t have if Google goes through with their foolish “enhancement.”

    • Martin Röttgerding

      Hey Mel,
      But isn’t it alltogther irrelevant if CPA for close variants is higher or lower? I mean, some keywords have higher CPA’s than others. If I find a new converting keyword with a 35% higher CPA then I’ll keep it – but I’ll probably lower its bid.

      The problem with close variants usually isn’t that they cover the wrong queries. The problem is that you can’t bid on them correctly.

      • Sam Owen

        It’s relevant in the sense that our choice is being taken away. If CPA is higher for them, we’re now going to have even more of our bids be “set” incorrectly.

        It’d be nice to have a close variants bid adjuster, but given Tablet isn’t an option yet, I don’t see it coming.

        • Melissa Mackey

          Exactly. A modifier for close variants would be awesome! Then I could still bid on them, but with the right bid. Same for Tablets. Same for Search Partners. I love the concept of modifiers, if only Google would use them where we need them…

  • Lex Urquhart

    Even when my company has disabled close variants, our search terms reports have always included close variants as match types. It may have been less, but it always seemed to me, and especially recently, that excluding close variants never 100% worked. Not sure if this was Google testing close variants on our own accounts, but I would have loved to have seen one search term report with no close variants (provided it also had sufficient data to not be coincidence). I guess now that they are removed, this is a pipe dream.

    • Martin Röttgerding

      Just to be sure: Are you aware that broad matches (including modified ones) also cover close variants? The ones you see probably come from broad match keywords.

      • Lex Urquhart

        Yes. I figured this was a given. We break up our ad groups by match type, and we receive close variants even when excluded on exact and phrase.

  • Martin Röttgerding

    Hey Sam,
    In your analysis you compared having close variants with not having them – but isn’t that the wrong comparison? I mean, the thing that’s changed isn’t that close variants were introduced. The change is the loss of the option.

    You wrote: “Two improved, one was about even and one was much worse.”
    Having the option to turn off close variants you would keep them for the first two clients and maybe the third while turning them off for the last one.

    Losing the option, there is no change for the first three accounts but the last one is worse off. Neutral for some accounts, bad for others. That’s it.

    All in all advertisers are either unaffected or worse off. We don’t have to be all dramatic about this, but the simple fact is that there is no upside to this.

    • Sam Owen

      Hey Martin,

      You are correct in that I’m looking at things backwards. However, this is because close variants are on for almost all Hanapin clients and I still feel the data is meaningful for showing whether they should be “on” or “off”.

      As I mentioned above, the need for drama is pretty minimal owing to the fact that they are typically “on” for us, it just seems unnecessary to force them on.

    • http://adclickbooster.de/ Steven Ruesch

      I think Martin’s view just about sums it up best. Taking away an option is just bad for those thereby negatively affected.

      One detail missing in the above analysis, as brilliant as it is, is that it was probably using data of accounts targeting English speaking users; however; (this is a big HOWEVER) Google’s performance in regards to close variants isn’t the same for all languages. In German I know of some examples where Google “thinks” two words are similar because root words within them are the same, but they represent two very different things.

      Losing the option to turn this feature off would result in them wasting at least 50-60% of their advertising dollars, even if they use the exact match keyword option. That is unacceptable considering that Germany is the 3rd or 4th biggest market for Google.

      God knows how Google performs for languages of smaller markets.

      While we can’t do much to avoid Google from blatantly using their position to earn an extra buck, it will in the long run backfire. The money earned here will probably be spent in PR campaigns in a few years time.

  • http://www.rankhammer.com/ Steve Hammer

    Sam – what’s the branding effect that you’re seeing? In a bit of cursory analysis that I’ve done, the percentage of non-brand traffic is a pretty strong factor in determining the result. Even with a strong brand, people will still mistype and still have strong intent. With non-brand the intent is far more varied and thus near exact seems to cause more degradation. Curious as to if you’re seeing the same, and if that’s part of the explanation behind the lead gen example above.

    • Sam Owen

      Hey Steve, that’s actually a great point and one which I wish I had more time to dive into. Especially when you look at the CPCs in my tables you can see clearly something is being impacted by branding. I’d love to read a case study if you can put one together from your data.

  • Larry Kim

    my own internal analysis came up with the number of 11%. not to far off from yours mike.

    • http://www.spyfu.com Mike Roberts

      Awesome. Thanks Larry. I got about $1.5M of new data in, and it’s all consistent, too. What were you seeing with Exact Match close variances? My data is largely saying it improves performance.

  • http://www.taptap.mobi/charts Jen Brannstrom

    In account #2 the table shows CPA on ‘exact’ rising from $57 to $371! That’s what I call a business-ending event. Or if you have other traffic sources, it’s high-time to switch.

    • Sam Owen

      Yeah it’s a pretty brutal increase! I wonder if there’s some brand data I need to account for, but either way, not good.

      • http://www.taptap.mobi/charts Jen Brannstrom

        “brutal” is the right word here

  • David

    Excellent post with solid examples, Sam. This can be filed in the ‘hey, we’re just trying to make it easier for advertisers’ spin folder for Google. Right next to the folder that says ‘how can we continue to extract more revenue for our stockholders from our advertisers’.

    Potential google adwords changes we can expect for the future:
    * keyword targeting is out. now we bid solely on customer affinities and demographics. bids for income levels > $100,000 automatically start at $25/click.
    * interstitial ads between google search queries and results
    * one click purchase from the search results pages (wait, i would probably actually pay for that)
    * SMS retargeting
    * no more options for remarketing. We’ll re-show your ad when and where we see fit. which will be to everyone, constantly. without end.

  • Sam Owen

    Thanks for sharing the link Julie!