• http://www.linkedin.com/in/jasonmanion/ Jason Manion

    The “click velocity” idea is interesting. Am I missing something, or is there a very simple explanation?

    The Google Adwords cookie only tracks conversions within 30 days, right? So in the example of the product that only gets 2 clicks a month, it would have to have a 50% conversion rate to ever see a tracked conversion. I don’t know about you, but 50% is a bit higher than most retail conversion rates I’ve seen.

    So these really low volume products might not be bad performers, you just never get enough clicks to really say. But when one does convert, it could pay for months of clicks on that particular product.

    I understand the concern about “death by a thousand cuts”, and I’ve seen that happen, but with no minimum bid in Google Shopping, I don’t see a compelling reason to cut those really low volume products.

  • http://www.thefind.com/ Shashikant Khandelwal

    If products with 1-conversion have an average spend of $22, that means only high-price, high-margin products can fall in this bucket. This spend cannot justify putting up low price products (e.g. baby bath toys) or low margin products (e.g. some electronics items)

    And even if the ROAS on 1-click products is 7.3, the total ROAS tanks like crazy when you add the $2.1M of zero ROAS spend in the denominator. After all, its the total ROAS on the campaign that matters.

    It seems to me that conversion-rate is not justifying the spend (i.e. CPC value) on low-converting products. So either the CPCs have to decrease (less revenue for google, good for consumer) or quality/coverage has to decrease (bad for the consumer, more revenue for google)

  • http://www.davidnrothwell.com/ David Rothwell

    If you see many-per-click conversions (old term) then visitors are buying more than one item.

    You may very well find someone coming for a low value item that upsells into something bigger.

    So it’s not always sensible to restrict your ads to a subset since you may miss these upsell opportunities.

    If only we had Search Funnels in PLA’s …

  • Lucas von Fürstenberg

    I would suspect they aggregated the data over the whole three months. So in your example you would actually have 6 clicks and no conversion.

    You still wouldn’t have enough data to say whether that product is worth keeping or dropping, but in that sense the click velocity makes sense as an indicator. You could argue of course, that all those “slow” keywords might perform a lot better if you monitor them over a longer time period as they might only get 1 conversion per year, but with 24 clicks still at a very goog conversion rate.
    The 30 days don’t really have to do with that, as you don’t look at one cookie, but at the product target.

  • Pat Grady

    Long tail keywords “behave” the same way. Split out the 0 and 1 sale keywords, then analyze each apart, and the long tails with one sale look great. There is no way to analyze insufficient discrete data like this – so, get more data. Cross reference your long tail against competitors pricing… which is what G is doing, in determining your click velocity for you… wait, I just proved your point. Dagnabbit!

  • Pat Grady

    Careful with that free shipping offer you only ran in January, it shifted the curves (yes, plural). Point being, you’re essentially studying an auction, not how many licks it takes to get to the center of a tootsie pop.

  • http://www.thefind.com/ Shashikant Khandelwal

    Remember, btw, that product catalogs change a lot over seasons (e.g. apparel), and definitely over years (e.g. canon cameras). With 1 click per month, it’ll take a long time to figure out what works and what doesn’t, and by that time, your catalog would have changed.

    So its more like you’re constantly in the mode of testing these long tail keywords, and paying google for those tests (in this test – $2.1M).

  • Lucas von Fürstenberg

    You have a valid point. Product catalogs do change. However having a product that is not searched for often and not offered by many, makes sense in a conversion rate point of view. Question will be whether Google will actually display your product when you have little to no history for the exact matching query or rather your competitors (or your) similar product.
    It would be interesting to monitor the products with that 2.1 million spent three more months and see how many of those leave that group within that span. I admit quite a costly experiment if I am wrong ;-)