• http://twitter.com/ichristianr_ Christian Rubio

    Paragraph after the aggregated IS formula. You applied the incorrect values to Lost IS (budget), using values from the IS column.

    Agreed on the other comment regarding spend efficiency, but I give the author more credit. That kind of managerial thought is better left for another post. This is strictly about diagnostics, prior to discussion about whether or not to spend more based on other marketing metrics and objectives.

  • Ben Vigneron

    Maybe the scenario I am simulating here is unclear to you. More specifically, I am referring to those paid search campaigns which ARE capped due to insufficient daily budget. If the only thing you do is to uncap those campaigns, then you can expect stable returns while generating more volume, since you’re bidding on the same keywords at the same bids, triggering the same ads, redirecting to the same landing pages, and targeting the same audience (geo, language, etc..). This is a basic scenario and it seems to me that your comment refers to forecasting revenue growth using Lost IS Rank, i.e forecasting revenue growth for those campaigns NOT capped due to insufficient budget.

  • Ben Vigneron

    Good call, I meant: Aggregated Lost IS (budget) =120,457*0% +58,789*12% +78,456*0%) /(120,457 +58,789 +78,456) =2.7%

    As for your comment about my conclusion, I totally understand that this analysis might not always be comprehensive enough to actually help make decisions in terms of budget re-allocation by market. There are so many potential variables involved, for instance the online to offline purchase rate or the return/cancellation rates might vary by country, in which case search marketers need to layer offline data on top of online data.

    However, for the sake of keeping this post simple, I had to limit the number of specific assumptions and externalities. Then you can make this analysis more sophisticated by adding your own assumptions based off other marketing metrics and objectives.

  • Nigel Haig

    Makes sense though I agree with Sam M that the far bigger challenge is forecasting using lost IM (rank), given you’ll likely have to account for higher CPCs etc. I look forward to that article.

    Regarding the debate on the assumptions used, I’d suggest it depends on how quickly your budget is being spent and your bidding options in terms of standard or accelerated. If you’re using standard delivery and you’ve enough budget for it to work then it’s perfectly reasonable to assume the performance levels will be the same. However if you’re using accelerated then it may be unwise to assume the same level of performance for the hours after the budget has been spent (likely to be at the end of the day), given user behaviour and competition levels are likely to be different at this time.

  • http://www.facebook.com/lucas.ashland Lucas Ashland

    To find Aggregated IS, wouldn’t you want to use Impressions Available instead of Impressions Received? Using Impressions Available would give more weight to the campaigns with a low IS and correctly calculate the amount of room for increased spend.

  • http://twitter.com/bennyblum Benny Blum

    It’s too bad IS metrics are set to be retired in AdWords beginning on Feb 4.

  • http://www.facebook.com/lucas.ashland Lucas Ashland

    Only the old IS columns will be retired. The new columns will still be available.

  • Pat Grady

    Loved your article, but the complexities of multi-touch attribution are sure to muck up your otherwise clean math and reasoning. We both know you can’t throw in the logic towel because it’s complex, and mentioning what the attribution model affects are, and how the may vary by country, is beyond the scope here. Not criticizing, wondering if folks will decide to allocate more budget to brand traffic, taking it from “lower” ROAS product ads. :-)