It’s not unusual for companies to ask us to take over the running of existing search advertising campaigns. This might be because a company has been managing things in-house, but wants to expand the campaign and feels it needs more time or experience than it has available. When we look at existing campaigns, if there’s a frequently identifiable problem, it’s not having kept control of the search terms where the ads are showing.
Even if you use strictly “exact match” terms where you want your ads to show, Google AdWords will take small liberties with your request. But the real problems occur with the occasional “broad match”, “modified broad match” or “phrase match” keywords. Your ads can end up showing in undreamed-of places if you’re not careful with these.
One example I’ve used in seminars was from an industrial control systems company which supplied components for CANbus systems. There’s a massively long tail of one-off searches occurring with “canbus” or “can bus” in them, so the company had specified “can bus” as a phrase match keyword in its advertising. Any search with that included was fair game, it reckoned.
However, when we looked, we saw that the advert had been appearing hundreds of times for searches like “can bus passes be used on croydon trams”. Now, you might say it’s pay-per-click, if the search and advert don’t match, nobody will click on it, and it won’t cost anything. That’s largely true, but some people will. More importantly though, the clickthrough rate for the advert was tiny (because it was showing on largely irrelevant searches), which results in a low “Quality Score” from Google and a much-elevated cost per click, even for relevant searches.
So how do we keep on top of this? I would suggest that in engineering or scientific search advertising, if any of your keywords are showing a clickthrough rate of under 1–2% in an average top 3 position, then there may be something amiss. The key is the “Search Terms” report in the “Keywords” section, which shows you the actual searches where your advert displayed. Or at least, it shows you the actual searches where your advert displayed and was clicked. All the times that your advert showed for other searches and was not clicked (the ones which damage your Quality Score) are not revealed. They’re hidden at the bottom under a label of “Other Search Terms”.
At this point, you need to decide which path to go down. If you’ve been doing this for a long time, you may find that the list of good actual search terms used is quite enough for your budget, so you can ditch the broad or phrase match stuff, and just work with a new list of only exact match terms, derived from the actual searches. If you still want to use broad match, it’s time to use all that data to compile an extensive “negative keyword” list. Just look for words or phrases which always indicate a search which you do not want to appear for. In the example above, we could probably set “passes” as such a negative keyword, so we appear for searches with “can bus” in them, but not “can bus passes”.