We use cookies to personalize content, to provide social media features and to analyze our traffic. We also share information about your use of our site with our social media, advertising and analytics partners. For information on how to change your cookie settings, please see our Privacy policy. Otherwise, if you agree to our use of cookies, please continue to use our website.

There Are No Bad Keywords

Raise your hand if you've heard someone advise advertisers to "turn off the poor performing keywords", or say "That group of keywords didn't work". These statements betray a deep misunderstanding of a fundamental truth in PPC: there are no bad keywords!

Many many attributes of a PPC ad can be bad. An ad can have lousy copy, a poorly chosen landing page, be targeted globally when it shouldn't be, etc., but the keyword itself is almost never a problem. This may sound like heresy. After all, I myself have written volumes on the importance of keywords, and the dangers of automated keyword generation.

So, how could I say that keywords can't be bad?!? Simple: keywords fire ads that generate traffic. The more targeted a keyword is to the advertiser's product or service and the more indicative the keyword is of buying intent the higher the quality of that traffic. Poorly chosen keywords bring in low value traffic, but here's the point: the traffic rarely has NO value.

When the ROI of an ad is bad the problem isn't the quality of the traffic, it's how much the advertiser paid for that traffic. Any traffic is good traffic if the price is right; the price might be zero for really un-targeted keywords (KW "Brittney Spears Pics" for a Pet food company), but usually isn't.

Obviously, you can run into major cost overruns by launching poorly targeted keywords with "category average" bids. There's no question that learning what the traffic is worth can be expensive, and launching stupid keywords does not help the cause. But assuming that we're not talking about keyword mistakes, but simply keywords that are general, somewhat ambiguous, or indicative of research rather than purchase the quality of the traffic isn't zero, so neither should the bid be zero. {Okay, sure, if the value of the traffic is $0.03 per visit and you need a 20% cost to sales ratio, then zero is the right bid.}

Now, we might find that for Acme furniture company, the keyword "furniture" only returns $0.30 per click in sales, and given margin structures Acme can only afford $ 0.08 per click on that term. Further, we might find that an $0.08 bid with our best possible ad copy and landing page still generates an ad on page 32 of the SERP and generates roughly zero traffic.

Practically the same as turning it off. But it is NOT the same. First, when we start talking about wide swaths of keywords rather than one or two, the affordable traffic we're talking about isn't zero. Moreover, that $0.08 bid might not keep you on page 32 forever. As other advertisers "do the math" they may find the value of traffic on that keyword isn't great for them either.

As the market rationalizes that page 32 ad might find itself on page 1 at some point, generating plenty of traffic and, all importantly, at a price the advertiser can afford. Many many advertisers and agencies make the mistake of bidding based on sunk costs.

Suppose in the above example Acme started bidding $0.60 per click for "furniture". They generate tons of traffic and see that their Cost to Sales ratio is a disastrous 200%. The answer isn't to declare furniture a bad keyword, and it certainly isn't to pause or delete the phrase. The answer is to bid what the traffic is worth.

There are other techniques to improve the value of traffic on a general term. Using a combination of match-types, sniffing search logs for negative associations, restricting the syndication, culling out poor performing ips and improving landing pages; all of that should be done and not just for the "problem children". But the the folks who say that reasonably targeted keywords "don't work" don't understand the first principles of bid management.

Join the Discussion