The reason radio has such appeal for us "microwave chauvinists" is that all targeted (directed searches with beams) imply you have the resources to simultaneously send these beams to enormous numbers of suitable stars for astronomically significant periods of time in order to have even the remotest possibility of finding one who just happens to be listening in that direction while your're transmitting.

A broadcast approach, where you transmit omnidirectionally, fills a huge volume of space with signal and even if the radius of your search sphere is much smaller, the number of potential targets is much larger. Of course, broadcasting is extremely energy-inefficient (most of your signal is wasted), but it stills reaches the largest number of stars for the least energy and infrastructure expenditure. A broadcast signal that can be picked up 200 light years away would reach hundreds of thousands of stars. That makes more sense than hundreds of thousands of lasers or parabolic microwave beacons that can reach tens of thousands of light years.

In my opinion, the optimum technique would be something in between, say a wide beam (perhaps the angular width of the Milky Way) and trade off penetration for volume. This beam could then be directed at areas of high stellar density, such as the galactic disk or nucleus and star clusters, in order to maximize the possibility of a hit.

For the passive partner, of course, listening will always directed at specific targets. At our level of technical development, we can transmit a direct beam that will reach across the galaxy, but our broadcast beacon would not cover many stars. However, it is safe to say that ETIs will be ahead of us in that respect, and will probably have no trouble filling up a sphere of radius several hundred ly (the bare minimum that will make it worthwhile statistically) with audible signal. Of course, once an ETI target's location is identified, the laser becomes the communications instrument of choice.