Here are a few thoughts:
Hide the number. If you only display the phone number on a clickable event, then you can track the number of times the phone number is requested, and extrapolate the number of actual phone calls based on a percentage of that number, which also makes it easy for you to track the path (keywords, etc) to the click event. The downside is that you may put some people off by forcing them to make that extra click to get your phone number.
Get a new number. You can always do what the phone companies have been doing for years and get a metered number. Chances are that your local phone company does A/B testing with their phone books, putting version A out to half of the coverage area, and version B out to the rest. By putting unique phone numbers on each set of the ads, which forward to the real phone numbers, they are able to determine which version of each ad produces the best results. As this solution logs a checkmark every time the number is called, you now have an exact count of everyone who called the number, and based on the metered number provider, you may have recordings of the calls or a way to trackback to the keywords searched on to get to the phone number page.
Baseline it. You could also try to establish a baseline of calls, and try to estimate the effects that the traffic to your page is having on phone calls, but this isn’t an optimal solution, as there can be other reasons for a change in your call numbers (seasonality, news about your locality / business, etc) . As with any marketing campaign, the more you can measure the success, or conversions, the sooner you can know when you’re doing something right, when you’re doing something really wrong, and determine your next steps.
If you’d like, you can also take a look at my entry in the Marketingpilgrim.com contest on Improving your PPC conversion rate. Feel free to give it a comment or heck, if you’d like you can even Digg it.
Thanks,
Simon Heseltine
