Choosing Negative KeyWords in your AdWords Campaign

You can reduce substancially the cost of your campaigns in AdWords and the ppc if you use the proper tools to detect which keywords or locations are not performing well enough. But getting ideas on which negative keywords to chose is not always easy. Among many other options, Expert Data Miner offers a very interesting report to spot the wost performers and upgrade your exclusion list.The following screenshot describes one typical example that can exist:

The software will try any combination with the search phrases that were used to access your website (from the pay per click) in order to measure the impact of any exclusion. In the above example some variables can help EDM to measure if a search phrase was "good" or "bad". The average number of pages requested during a visit is an indicator, but not as much as the number of pages that are requested after an initial ppc hit or the percentage of rebound after such a hit. A high percentage of sessions with mixed requests (organic searches and ppc searches) can be considered as a bad thing, it suggest that many visitors visited you twice with the same search phrase, once with an organic search and once with a ppc search, so there is some redundancy here. The position of your website on Google for some keywords is so good that many visitors would find you with an organic search anyway, so you probably paid in vain. The samething can be said about the last column, the percentage of organic hits for any search phrase that uses the combination of keywords that you find in the first column.

But the most valuable indicator is the conversion rate. You just need to build a custom column in the report, a column that gives the percentage of the visitors who will ask for a specific page during a visit, like "/thank_you_for_purchasing.html". Ideally the page should be chosen to get a conversion rate between 1 and 10% to have the proper statistical sample. Whether your visitor confirmed the transaction or decided to buy by phone after is irrelevant; you don't need to have the real conversion rate but simply a conversion rate that is strongly correlated with the real one. If 3.57% of all the PPC visitors put something in their cart (as here) we don't need to know if only half of them confirmed the transaction or if 20% of them were scared to give their cc info, changed their mind and decided to complete the transaction by phone. All we need to know is that if, for some keywords, the "conversion rate" is 1.2% instead of 3.57%, it's likely that you also sell 3 times less with those keywords, provided that your sample is big enough to draw valid conclusions.

In the above example some keywords like apartment mailboxes or box cadeau perform badly. The bigger your sample is, the more you can take the correct decision.

If you click on any combination of keywords in the first column, you fetch a sub page that contains all the occurences for that combination, al the paid search phrases that were used and the number of hits for each, as well as their conversion rate. The sub page looks pretty much like the main page, except that it is dealing with only one combination of keywords. In order to get this report, you just need to scan your log files with Expert Data Miner, go in the report on Search Strings (phrases), and click on F6 to get the click fraud report. The report on negative keywords will be given when you exit the click fraud option - or when you cancel it at an early stage -.

Back to the home page

Download Expert Data Miner