Forum Moderators: open
- up near the top:
4589 washington hotel
- then, all sorts of variations on the phrase,
and, further the list:
102 washington hotel
I'm seeing this a lot... I've even seen one spot where the phrase is repeated 3 times. It might be singular/plural, or different variations on word order made to match MatchDriver and not combined into one.
Usually, though, when they mess with word order they go alphabetical. I'm seeing word order preserved enough that I'm seeing entries like this:
146 hotel washington washington dc
And in the past, they've always just combined singulars and plurals. Any thoughts?
If "search terms for september" gives different numbers for the same term at different times (and grossly different numbers), at what point is that considered deliberately presenting misleading information to increase profit?
I'm just adding the two up because I don't take too much stock in the figures anyway, especially on higher competition terms that could have a lot of position checking done. I only use the figures as a comparative basis for relative worth of keywords, not as an absolute.
August:
- querying "repairing widgets"
6000 repairing widget
- querying "how to repair widgets"
3000 repair widget
September:
- querying "repairing widgets"
450 repairing widget
- querying "how to repair widgets"
100 repair widget
Checking on Overture itself, this is very non-competitive area, only one bid on the top phrase, for .09. I hadn't checked whether there was bidding on this in August.
I have a little bit of egg on my face over this one. I'd advised a prospect (now a client) that smart targeting was the foundation of what we're doing... and even though I refused to make click-through predictions, citing the unreliability of Overture data, I also cited "widget repairing" as a better target than what they were targeting, which it no longer is.
I've never experienced a jump quite this big.
If other terms were down proportionately, I'd say that would make sense. But looking at two other phrases, which last month didn't look like such good targets, they've both actually gone up slightly... and they both now look better than terms I'd suggested:
August:
450 widget fixer
950 fixing widget
September:
550 widget fixer
1100 fixing widget
WordTracker is showing numbers so much smaller overall, with enough 5-word strings searched a bunch of times, that I'm doubting their numbers... which I think are extrapolated up besides. It would be a wonderful tool if they had a larger database. They do show that "widget repairing" and "repairing widget" combined are searched more than "widget fixer," but not by much.
Google AdWords Select had "widget repairing," "repairing widget," and "widget fixer" each with about the same number of click-throughs, though that's also a function of targeting... not just searches.
But, to get back on topic, on Overture there was either a big anomalous bubble in some terms last month, or a big dip this month... as well as the multiple entries we're all seeing.
August:
- querying "repairing widgets"
6000 repairing widget
September:
- querying "repairing widgets"
450 repairing widget
Not quite so. For September, querying "repairing widgets," is actually reporting:
450 widget repairing
- No searches were reported at all in September for "repairing widget."
- In August, they'd reported nothing for "widget repairing."
(I'm so used to recent Overture results blending all word orders into one that I didn't notice the distinction).
So, I'm theorizing that these little subgroups of numbers we're seeing might be the phrases searched in various ways and returned with word order re-arranged by MatchDriver. And then maybe... as they're ironing out the system... something went awry that affected the results I'm looking at and they dropped a whole group.
?-?-?-?
I think you can get a reasonable idea on the competitiveness of certain keywords by using BOTH Overture and Wordtracker. But I'll reiterate what Marcia stated earlier that neither is an absolute and keep in mind that Overture naturally creates an environment where the very competitive keywords have artificially inflated results.
