• http://www.mynextcustomer.com George Revutsky

    There seem to be 2 big implications to live updating results, that most folks I talk to seem to overlook.

    1) This is, potentially a huge jump in revenue for Google due to the pre-emption of long-tail queries

    2) This may have a significant negative effect on the larger, more sophisticated paid search advertisers who have a lot of long tail keywords and log account histories.

    Here’s why I think the above – would love to hear other perspectives.

    When a searcher’s 5-6 word query is frequently pre-empted by a 2,3, or 4-word suggested query, the resulting paid search ads will cost more, on a per click basis, than if the searcher completed their original (potentially longer-tail query).

    In general, 2,3, and 4-word phrases are generally more expensive.

    For example, along comes a searcher who is looking for “cosmetic dentist in walnut creek specializing in Britesmile”

    They start to type, and the suggestion bar and the live-updating results pre-empt them completing the search and they choose to just search on “cosmetic dentist”. VOILA! Google just got to serve up ads against a shorter, more expensive phrase. More revenue.

    Yes, this adverse effect does partially depend on the keyword match types in the dentist’s account. If they weren’t set up correctly in the first place, both the long tail and shorter search might have been priced the same. But if they were, it would hurt them.

    2) SERP Load Acceleration = Revenue Acceleration

    Yes – searchers will save a little bit of time typing. And this will accelerate the time from search to click. So more revenue – a little bit faster each day.

    If you take a few milliseconds, maybe even a third of a second, saved on, say, 5-10% of all searches done by Google, that’s a LOT of acceleration and quicker results (as they are pre-loaded).

    3) A Wealth of New Searcher Intent Data
    By seeing what people start to type and what they end up settling on when presented with essentially a multiple choice set of options (a, b, c, or “none – let me complete my query”), Google gets an amazing ability to model what kind of phrases can pre-empt long tail searches, and what kind actually trigger them.

    And then they can see what kind of people this works with and what kind it does not.
    My guess is, you can build some very different types of models in such a situation, rather than simply analyzing what is being typed in across a huge volume.

    For a simple example, you can look at gender differences across verticals when faced with what amounts to multiple-choice prompts.

    Some studies, for example at Sweden’s Goteborg University, show significant differences in male vs. female performance on multiple choice and free-response formats. (http://bit.ly/d6n5m3) The results of that study show that women generally perform better on free response tests, while men generally perform better on multiple-choice tests.

    Now, I have no idea if that is really the case. Please let’s not start a gender aptitude conversation here – that’s not my point at all.

    I just mean that if you’re Google, you can start to make inferences about demographics vis a vis the autosuggest field. You can see how folks interact differently with the auto-suggest, and, if someone is logged in, change the autosuggest for them. Gender could be one of many things you look at. Same with age. Or geography. Or maybe none of these things.

    Finally, you can even look at prior behaviour, which is considered by most database marketers to be more predictive than demographics, and tailor the autosuggest to THAT.

    For example, if I went to 3 adoption agency websites in the recent past, then started a search with the phrase “adoption…..” Google couldmake most of its auto-suggests be about the adoption of children, and maybe only 2 will be about pet adoption.

    The potential of this tool is mind-boggling actually.

    To sum up, the whole point of this update is to a) make the Search query about more expensive keywords b) load the results faster and c) make the searcher’s likelier to click on a paid result.

    Guys – I think this update is a really, really big deal. Would love to hear everyone’s thoughts. Would have posted this on out blog, but we’re just getting our new one up.