Larry Kim recently wrote a good article for Search Engine Watch on a subject that's rarely talked about in PPC land - advertisers break out their keyword structures into super-granular themes as best practice. However, after setting live, they leave these keywords in their accounts regardless of performance.
Why is this bad?
- It shows you're not taking learnings from your activity.
- Exact match keywords that are too long tail and flagged as 'low search volume' by Google aren't eligible to show in the results, so if you've applied cross match negatives for these keywords to your broad match modified (BMM) campaigns, then you're not eligible to trigger this term at all, limiting your visibility.
- Having tens of thousands of keywords in your account (or hundreds of thousands, or even millions!), makes for a messy account that's tough to navigate. If you have a huge account you really should ensure you're able to take learnings from all of these keywords you have live, else you're just wasting your money.
- In addition to the above, poor performing and non-performing keywords brings down the overall Quality Score of your account, inflating the bids you're paying on keywords that are important to you.
I'm encountering the above problems, what do I do?
- Run a filter on your keywords and filter out all keywords that have 0 impressions over the last three to six months and troubleshoot why they're not gaining impressions. If they're 'low search volume' or keywords that have traditionally performed badly, and you have no interest in running them again in the near future, delete them.
- Filter to your poorest QS keywords (scores of 1-3), and troubleshoot why they're so bad, again if you have no intention of investing in these keywords going forward, remove them from your account.
- For all you've removed, ensure you've deleted the corresponding cross-match negative from the BMM campaigns so that if those search terms do increase in search volume, your BMMs will pick them up and you'll be able to see their performance in SQRs (and if it's actually worth breaking out into their own exact match campaign).
