bid management

How often should you check your bids?

Working with Manual Bid Adjustment Formulas

Hey Roy, I was manually setting some bids using your AOV * CR * Target COS formula to generate a bid for some pay-per-click display advertising campaigns.  I think I must have used too short of a window when looking back at the AOV and conversion rate because it kicked out a bid of $1.52 and within a weekend the campaign had blown a ton of dollars and generated a 148.27% COS.  I re-ran the formula using an updated trailing 30 day data set and now the formula kicks out a $.07 bid.

Here is my thinking, one of the things that someone probably should look at when running the formula is the stability of AOV and CR.  There are probably some good rules of thumb if there is a bunch of variance in those numbers.  So, what's your advice?

A longer window is the best protection against that kind of swing.  Don't be afraid to use 60, or even 90 days, when data is either scarce, or volatile.  That's like making your kayak longer, before you head into the waves.

The frequency of your updates should be less than a quarter of your window length.  If your using 30 days, don't let it go more than a week.  With a longer window, changes will be more gradual, and you can react a bit more leisurely.

So, based on what it sounds like is going on, use the size of change as an indicator of whether the volatility is too high, and use a longer window to smooth things out.

Make sense?

Have questions?  Drop me a line via the Contact form--I love to talk shop.

Up and to the right!

Should we use the AdWords interface or a bid management tool?

What is your general preference for managing PPC - in the AdWords interface or using a bid management tool?

My automation strategy has always been "do everything manually, and automate the biggest pain in the ass first." It's served me well for years. I recommend you do the same. You'll learn about all the ins and outs of your account, campaigns, and performance. This will arm you to be better equipped to either automate your process, or evaluate third party providers. The worst thing would be to hire someone, and just take their word for it. Never let a third party tell you how well they're doing--measure everything independently, and that's something you'll need to learn how to do. There are three layers of options for doing it yourself. First, manually. Second, AdWords Scripts, for automating most of the things you were doing by hand. Third, the API--it's more complicated than either of the first options, but has a lot fewer limitations. Still, due to Pareto's Law, I do almost everything with the first two.

Have questions?  Drop me a line via the Contact form--I love to talk shop.

Up and to the right!

Is day-parting worthwhile?

Roy, we'd like to know if we should use day-parting?

Maybe, but not usually. If you have some behavioral explanation for why conversion rate would be different at different times or days (not volume, it's never a traffic question, aside from "do I have enough to reach significance?"), then perhaps. In most cases, changes over time result in differences in volume, but not conversion rate, and therefore no change in bid. You might see some shifts over time, though, and you can tell if they matter by using the same math as I described in the Geoparting answer. If you have something worth bidding after, you'll find it that way. If you're not sure, leave it alone--you're more likely to making things more complicated than they need to be than you are to optimize anything by hand.

Have questions?  Drop me a line via the Contact form--I love to talk shop.

Up and to the right!

When should we use geo-parting?

Is Geoparting worthwhile and when should we use it?

Here's how you can tell--take the state that has the largest number of clicks, and the state with the largest deviation from your overall conversion rate. Check their conversion rates against the overall, just like you would with an A/B test (I recommend the two-proportion test in WolframAlpha). If the certainty is better than 95% for either of them, you should take a look at all of them, and make proportional adjustments where there are significant differences from the overall, and leave the rest alone.

Have questions?  Drop me a line via the Contact form--I love to talk shop.

Up and to the right!

Making mobile bid adjustments

Hey Roy, when should you make mobile bid adjustments?

Mobile bid adjustments are easy and it's one of the most common errors I find when doing account reviews. Making the best bidding decision for traffic coming from mobile devices is a function of the difference between your mobile and overall conversion rates. If your mobile conversion rate is the same as your overall, then adjust by zero--the ROAS target you're aiming for is being satisfied, after all. If your mobile conversion rate is 50% that of your non-mobile, then drop those bad boys by half. Rinse repeat every month or so, to account for shifts in behavior.

Have questions? Drop me a line via the Contact form--I love to talk shop!