Why Most Bid Automation Packages Suck

About The Author

Dennis Yu is the author of this guest post. Dennis is the CEO of BlitzLocal, a localized internet marketing company. He is an amazingly smart and talented internet marketer. Also, thanks to Nick Abramovic, who assisted with concepts in this article

Why Most Bid Automation Packages Suck

At SMX Sydney last month, I spoke on the Expert PPC panel. There were several web analytics vendors (I won’t name them here) who all but claimed that their proprietary bid optimization software would eliminate the need for you as a PPC marketer to do anything. Just watch the cash roll in. How does this software actually work? Well, you’d have to be a Ph. D in Statistics to understand– but drill down and you’ll find that it’s really not that difficult. No bid management system we know can beat the power of a motivated human, who can account for brand terms, seasonality, competitive actions, what’s going on in TV land, or any of a zillion factors that aren’t quantifiable by a change in CPC, average position, and conversion rate.

Agencies scale not by creativity but by overhead. This is true even with agencies who use either in-house technology or license technology from others. Even the advanced agencies are not very sophisticated – they will use rules-based bidding that works half the time (you still need a human to double-check) and because they are rules-based – they require you to make customizations (or as they say “settings changes”) depending on your account/campaign. We’ve taken over a couple dozen campaigns from the big boys and when we look under the hood, it’s still hampsters on wheels, not technology. If you’re a small affiliate– a one-person show, take heart to know you’re not at a disadvantage! The emperor has no clothes.

What are some issues with rules-based bidding? Not enough data. So you’ve got 10,000 keywords x 10 positions x 10 ad variations x 10 landing page variations x 3 search engines. That’s 30 million combos. And then the supposed bid automation tool vendor says they can manage those 30 million combos each day for you in real time. Why, certainly, you wouldn’t want to manually try to look at that, would you? First off, just like the density of matter in outer space, 29.95 million of those combinations are likely to get you zero impressions, not to mention clicks or conversions. How is the tool to make a determination if there’s no data?

What if Keyword A gets 10 clicks and 1 conversion, while Keyword B gets 8 clicks and 0 conversions? Can you tell if one term is better than another? I’ve actually interviewed someone who claimed they can determine if a keyword would convert in just 10 clicks. Reminds me of the old TV show “Name That Tune”, where contestants compete to see how many notes they can name a tune in. While most contestants could get it in 5 or 6 notes, the reality with PPC is that required sample sizes are dependent upon factors such as confidence intervals and expected conversion rates. If an offer converts at 1 in 100 clicks on average, you can’t tell with just 10 clicks.

So let’s say you run the calculation and determine that you need 200 clicks to make a statistically significant decision. How many keywords meet this threshhold? What if you split those measly clicks further into time of day, destination url, and other factors? Pretty soon you’ve got but a handful of clicks to make a decision, no matter how nice your Taguchi tests are set up. The solution– set up keywords into tightly themed ad groups. Make sure that keywords in an ad group mean the same thing and are equally likely to convert. Then you can bid at the ad group level.

What bid management software will group your keywords properly? None, unless there is one that can write and speak as well as you– Alan Turing, we’re a long ways off from that. So don’t let magical software substitute for good old fashioned thinking. Use affiliate PPC tools appropriately. Landing pages are 10 times more important than ads, which are 10 times more important than keywords, which are 10 times more important than targeting, which are 10 times more important than bid optimization.

At one conference, I met an advertiser who boasted of having 3 million keywords in his portfolio. He wanted to get advice on how to manage such a complex campaign. I asked how many of those terms got at least one click in the last 14 days. About 15,000, he said. Get rid of the other 2.99 million keywords and you’ll have an account that’s easier to manage and will make you more money. His campaigns were complicated and crappy– at least take the step to make them simple and crappy, where you stand a chance. No amount of cosmetic surgery will fix it.

You can’t make chicken salad out of chicken shit. So a great bid optimization package won’t fix a poorly built campaign. Got extra cycles? Work on your landing pages and ad copy. Want to send competitors on a wild goose chase? Go on and on about how bid management through some magical software or other set of tools has been so key to your success.