Bid Management Automation: Three More Inconvenient Truths

Last time, I covered some shaky assumptions behind bid management automation, including the assumption of an easily pinpointed target CPA. In this column, I’ll address some additional issues that don’t often come to light in industry discussions. While they may make some vendors squirm, they also might help advertisers select the tools that are just right for their needs.

Three Inconvenient Truths

1. Frequent bid changes through the API cost money. Costs may vary, but frequent changes require you to pay the search engine for API tokens used. Bid management solutions generally set you back something like 2% of your ad spend.

At least one such tool, though, is absolutely free. If you like free, consider using Google’s free built-in Conversion Optimizer. If you have a threshold level of conversions per month in your campaign, Google Conversion Optimizer will attempt to keep your campaign within a specified CPA range. Clearly, to do a decent job on limited data, such a tool would do well if it could interpolate bid strategies for like keywords and like scenarios, based on the limited data it does have, rather than waiting for statistically significant data to build up on all keywords.

No company is better positioned to make such interpolations than Google: they simply have more data and more development resources than any third party. Strategically, though, Google’s tool has all of the same shortcomings outlined in last month’s column.

2. Google’s system can provide greater accuracy based on Google data collected. Let’s say you have four ads running in an ad group. With its mountains of data, Google could conceivably decide that one ad is better with some of your keywords, and another ad goes better with others. Google has never made explicit guarantees that ads will be fully delivered based on a set of rules you can literally follow and reverse-engineer. Quality Score is increasingly complex; your ad may be deemed ineligible to show up at any given time, but in this dynamic environment, that might only be the tip of the iceberg.

If Google’s own systems are trying to optimize your ad delivery to connect keywords and ads with the most engaged consumers at the right places at the right times, then arguably, again, all you can do is set a bid with the expectation that the “correct” bid could fall within a wide range. Google’s systems today or in the future could play around with the specifics of ad delivery, position, etc., mainly because Google has every intention of aggressively determining the Quality Score of every keyword in real time at the time of the query, and those scores affect ad position, sometimes dramatically. In such an environment, third-party bid management systems are selling you false precision.

3. Delivery control (or budget allocation mechanisms) might be superior, strategically, to bid management. Why? Because the way “bid management” saves you money is to lower your ad position — sometimes, to a position of near invisibility. (Actually, it’s even worse than that. Many systems are all fired up to *pause or delete* keywords at the first hint of non-performance). Bid management can also improve profitability by taking the obvious step of placing high-ROI keywords in higher ad positions. But that’s not the only way to “manage” a keyword portfolio to run long term tests of long tail keyword viability.

There are currently only one or two very blunt mechanisms to help you control the proportion of the time that your ads show for your keywords. The primary one is the daily budget. Google limits your ad impressions campaign-wide if it looks like you’re going to exceed your daily budget.

Most spending control mechanisms are bid-based. The advanced dayparting (Ad Scheduler) feature in AdWords, for example, allows you to bid lower at different times of day. But bidding less impacts your ad position, and that combined with lower CPC’s is the mechanism by which you expect to spend less at those times of day on that campaign. What if you wanted to control delivery by simply limiting delivery, in a way that you had more control over, without using ad position and CPC’s as a way to cut that spend?

Take a given long-tail keyword that has not performed to date: ask the system to keep it in play, but only at a 15% delivery rate. Its potential drain on your account performance is slashed dramatically, but the keyword stays in play. The “test” keeps running. The alternative, turning the keyword off or bidding lower in order to drop its ad position to somewhere ludicrously low like 24th, isn’t any better than delivery control at achieving your budget allocation and testing objectives, and may be worse. Similarly, with advanced dayparting, wouldn’t it be nice to use delivery control to manage your suspect times of day, as opposed to plunging your keywords down to weird low ad positions that behave entirely differently?

Unfortunately, some of these pipe dreams of mine (like 15% delivery on long tail, under-performing keywords) may conflict with point #2. Google may itself move towards a system that “gives up on” certain words in your account by assigning them prohibitively poor Quality Scores. It might even consider using the ad delivery trick by giving them a “slight chance” to show up in visible ad positions, but only say 15% of the time. I think it would be more fun if advertisers themselves (or their bid management systems making such determinations through the AdWords API) had that kind of control.

Improved strategy prevents future headaches

In any case, by taking a frank look at some of these issues, we at least move closer to being able to develop useful strategies for large accounts. Turning to a kind of technology that often goes hand in hand with the need for bid management automation: we haven’t seen the end of the development of “long tail keyword generation tools.” From what I am seeing, the next generation (even beyond the fairly ambitious current generation) of long tail keyword generation tools are all about “push”.

Push advertisers into accepting that they need an even longer list of low-frequency keywords, with no proper plan to handle those lists. What does the long tail keyword promoter care if his lists cause more trouble than they’re worth? To use the automotive analogy, we need a great suspension and solid handling if we’re going to add more horsepower to the engine.

In almost any human endeavor, you can point to a lot of variables that pump up the apparent complexity of the task at hand. If you want, you can compute all the angles taken by all of the bones in your feet as you walk to the corner store to buy milk. But the milk won’t be any cheaper when you arrive.

In that respect, many bid management vendors are selling false precision based on a premise of complexity. The claims are often misleading and under-represent the depth of strategic insight and experience required to manage an account. As I’ve argued, many of the solutions in the space do an adequate job of helping you handle the problem of account “bigness.” But we need to dig in a bit more to decide if apparently advanced systems are worth the cost.

Even considering this topic in the depth we have thus far is taking our focus away from more challenging, higher-order tasks involved in creatively reaching and engaging customers online. There’s an exceedingly smart computer resting above your shoulders. You’ll need it.

In next month’s column, I’ll follow up with guidelines for selecting a bid management vendor.