• http://www.cpcsearch.com Terry Whalen

    Hi Tom,

    I think your takeaways are solid, if underwhelming. Who would argue against the notion that there is more search query data than keyword data (that is basically what is being said: “there’s a longer tail for query data”)? This is true almost by definition, as many search queries are unique – there will always be more search queries than keywords, even when folks are diligently using search query data to add keywords to accounts. And I don’t know anyone that would argue whether negative keywords are a good idea. How could they not be? Of course they have value. And anyone that argues that the long tail has no value or is unimportant either doesn’t know what they’re talking about or is simply trying to come up with article copy.

    There’s a lot of value in the short tail and the long tail and the tail in between, and it’s certainly a good idea to try for additional granularity by making sure keywords map as much as possible to actual search queries that ads in ones account are being matched up against.

    But it’s also possible to go overboard on this – if your account has 100,000 keywords but only 5-20% of those are showing any cost data at all in a given month, you may want to actually prune keywords (and afterwards, add keywords for which there are actually searches being performed by actual searchers), so that your account becomes less unwieldy and a little easier to manage on a day-by-day basis.

    p.s. I am a fan of Wordstream!

  • http://www.rimmkaufman.com George Michie

    Hi Tom,

    Thanks for the kind references.

    Question: why are the costs per conversion different among the groups? Seems like if you’re willing to pay $X for a conversion that would apply — on average — as much to the long tail conversions as to the head terms. We’ve never gotten instructions from a client to the effect: ” I’m happy to spend 30% of my revenue on short KW, but don’t spend more than 10% on the long KW.”

    Might differences in the cost per conversion indicate faulty bidding, or is there some other explanation?

  • http://www.reidspice.com reidspice

    Interesting analysis. I have to say, though, this sentence raises some serious questions about the cleanliness of the results you’re presenting:

    “The interesting thing about this analysis is that a vast majority of the spend in the accounts in question was directed toward long tail search queries.”

    I’d say that’s not true of many (most?) paid search accounts.

  • http://www.tomdemers.com Tom Demers

    @Terry: Hey Terry! Hope all is well.
    Yeah I suppose the thing that I found interesting/compelling was the way the actual data broke down from a CPA/volume basis for these advertisers. The point of interest might be more the way that these SMBs are structuring their accounts, and the fall out in terms of distributed costs, conversions, and CPAs. The importance of search query data and the distinction between keywords and queries is obviously something you’re very familiar with, but for these advertisers not adequately managing for the distinction seems to have created some waste in certain areas.
    @George: no problem re: the kind references, I’m a huge fan of your blog and your SEL column!
    Yes bidding is definitely a component I overlooked in the article. As I said the data is a mix of new WordStream client data and more frequent users, which (among many other things) makes the data far from perfect, which I hope I helped to qualify in the article. Many of the clients are likely failing to manage bids effectively at the query level – particularly so because even in the instances where the data is representative of WordStream customers, our software isn’t a bidding platform.
    I suppose the take-away I didn’t adequately articulate is that these numbers are an interesting indication of what many “steady-state” paid search accounts look like for SMBs. I’m sure RKG and CPC Search client accounts look different.
    Anyway great comments guys: I’m a big fan of both of your writings whenever I see them around the Web.