Secret Truth Series #15: Bid Lipstick on a Keyword Pig

Bidding is the magic elixer of paid search. It’s the solution to any problem, the driver of all success and the cause of all failure. Once you have the right keywords, bidding is the main activity of paid search management. It’s critical to get your bids right. If your bids aren’t properly set, intelligently managed, and constantly cultivated it’s effectively impossible to succeed at paid search. This all is common knowledge. And it’s all untrue. The idea that bidding is important/critical/central to paid search is based on several assumptions:

  • We have good data on which to make bid decisions.
  • The data we have from the past is predictive of the future.
  • Our algorithms are able to take this good and predictive data and recommend a better bid
  • The effect of a bid change is clear and direct

Let’s look at those one at a time.

Lipstick On A Pig

Keywords are the false gods of paid search, as we discussed earlier, but they have another problem too: keyword metrics aren’t really reporting on the keyword itself, but rather on the combination of the keyword and all the options that are set around it. Click-through rates and conversion rates, for example, are the result of how targeted the search queries were based on the match type used and the quality score received, how appropriate the ad copy was for those queries, and how persuasive the landing page and offer were to the people who clicked through. That same keyword matched to different queries, with different ad copy, and different landing pages and offers would/could perform much differently. It’s shorthand to say that the keyword had a 2% conversion rate – it was really the entire advertising program built around the keyword that earned that result. Yet bid algorithms, with our approval, make bid decisions as if the keyword metrics were much more conclusive than they really are. Suppose the queries change (which they do every day) or the ad copy is rewritten or tested (which it should be several times within the 30 day window used for most bid decisions). Isn’t it kinda kookie to ignore these variations and just pretend that the data is clear or reliable? In ’21 Secrets’ I talk about getting keywords ‘bid ready’. The idea is to create a stable and functional environment around the keyword, to deliver better and more stable data, before putting much effort into bidding. As the book points out, worrying about bidding on keywords before they’re ‘bid-ready’ is a text-book case of garbage in, garbage out.

Past Results May Not Suggest Future Performance

So if our keywords aren’t bid-ready the data on our keywords is pretty dirty. And in most cases (probably 80% of active keywords) the data is also really sparce. There just aren’t many clicks or conversions on most keywords over a week or even month long period. Predicting the future based on sparce data about the past is extremely difficult. And even is there is plenty of data, it’s a pretty big assumption to think that last week is gonna be like next week. The world and our businesses don’t operate at a steady state. So even if we’re not in a strictly cyclical business – where summer dresses have a rise and fall – there are all kinds of time-based or external factors from competitor promotions to weather to holidays to news events. One only has to look at a performance trend report of any keyword – on CTR, conversion rate, even impressions – to see that if you picked any one period from the past and projected forward as if that period was representative, you would have misjudged a bunch of what in fact happened later. Yet that’s what the bid algo’s do. It’s hard not to believe numbers – as pointed out in the recent post on averages – especially when we’re working so hard to get them. But it’s not really that the numbers are lying, we’re just reading more into them then they can accurately provide.

The Smart Black Box or The Perfect Rube Goldberg

But forget all of that – take all the imperfect data, assume the past predicts the future in ways it probably doesn’t, decide which and how much data to consider (another massive black hole we’ll skip over), and hand it over to your bid algorithm for analysis and a recommendation. There are all kinds of bid algorithms. Some use simple math while others rely on very complex mathematics. There are also all bid rule systems (which differ from algorithms) and base decisions on a series of measured factors and nested If/Then/Else criteria. Few or none of these take into account that the data they’re based on may be flawed or even terrible. All of them ignore dozens or perhaps hundreds of variables and factors that almost anyone would agree clearly effect ‘the right bid’. Regardless, they do their best and out comes a new bid recommendation. Good luck with that.

Why Does The Garage Go Up When I Flush This Toilet?

What does raising (or lowering) your bid mean anyway? The relationship is not simple nor direct. We know how bid connects to Ad Rank (Secret Truth #9) and how Quality Score impacts bidding (Secret Truth #10). Together these explain that what we’re willing to pay (our Max CPC or Bid) is only indirectly related to our the amount we do pay (our actual or average CPC). And many other factors are at play:

  • In some cases, our actual cost-per-click is far below our bid. In other cases, our cost-per-click is consuming 100% of our bid.
  • If we change our bid but our position (ok, average position) doesn’t change, what effect did the bid change really have?
  • Quality score is unique to every keyword-text ad combo, changes per geography, is influenced by the query, and evolves over time.
  • Raising a keyword bid also expands the scope of matched queries for which you can enter the auction and earn sufficient ad rank to appear. So like butterfly wings changing your bid alters the environment you were trying to control or effect.

I could go on. And yet just about every bid algorithm I’m aware of ignores – all this and more. It’s hard to shake the notion that bid up = pay more = higher position, but strictly speaking it’s far from that simple.

The Twenty Percent Solution

So what is an advertiser to do? Two things: 1) Spend your time organizing and optimizing your campaigns and keywords before you worry much about bidding. As the 21 Truths in this series highlight, there are a lot of ways to impact your results that are directly within your control. Master them. . 2) Think about bidding as having *at most* a 10-20% impact on your success. Bids interact with keywords and match types and quality scores and text ads and landing pages and everything else. Treat them as part of the system. It was via Brad Geddes via David Szetela that I first heard the significance of bidding put into its place with these kind of numbers. I’m pretty sure I later saw Dennis Yu give a clever talk and example that drove home the same point. I’m sure others have said it as well. There is no way to say whether bidding in general a 5% factor or a 45% factor – it doesn’t matter and it certainly varies from keyword to keyword. But what is clear is that for even the best bid algorithms or rules or strategies to do their job, they must be applied to well designed, managed, and optimized campaigns and keywords.

  • A good bid on a bad keyword (or based on bad keyword data) leads to a bad result.
  • A good bid on a good keyword-system plays an important role and can absolutely make or save you money.

Bidding Is Important

This post wasn’t intended as bid-bashing. As the book says, I believe you should spend a lot of time on bidding after you’ve spent a lot of time on everything else. But bidding is very often over-emphasised, mis-represented, and mis-prioritized. The right way to bid, the right algorithms, the role of rules, the solution to sparce data, the implications of time, the way to factor in all the factors – these are all topics for another time and place. Until then, make sure your keywords are ready before you set or manipulate their bids. And have reasonable expectations for the bids and bid changes you do make. They’re doing the best they can, but the available information and state of the technology limits severely limit their potential. What Do You Think?

This blog post is part of a series extending and amplifying the ideas in our free ebook ’21 Secret Truths of High-Resolution PPC’. What they’re saying: “Everything you know about AdWords is the basics Google wanted you to know. Just enough to get you hooked. But what if there was fundamental secrets that they neglected to share? Would you want to know them? Now you can! 21 Secrets Truths is what you must read, no, act on, before your competitors do.” – Bryan Eisenberg Conversion Expert and New York Times Best-Selling Author ’.

Acquisio Admin

Acquisio Admin

The First Machine Learning Marketing Platform
Built to Scale Search for Local Resellers & Agencies

Automate, optimize and track more campaigns, more profitably.