In my technical areas of electronics and software, I’ve often heard in-house counsel say it isn’t worth the cost of doing patentability searches. They are trying for a large volume of patents, and it isn’t worth a lot of budget on each one. It usually goes something like this:

A patentability search can cost $1-3,000. That’s an additional $20,000 on average for each 10 applications. If they save you from filing two applications, it is only breaking even. In most cases, it just means you get narrower claims, so the application preparation costs aren’t saved. When the Examiner does his/her search, we’ll deal with the prior art then – that’s what dependent claims with fallback positions are for.

I’d like to suggest that this ignores a number of cost saving and quality benefits in addition to saving patent application preparation costs once in a while.

1. A patentability search can save the entire prosecution cost, not just application preparation costs. Without a patentability search, an application is not only prepared, but government fees are paid, and prosecution costs of multiple amendments are paid, until the claims are narrowed so much that they are abandoned as worthless or issued with narrow claims with little worth. Since it often takes extended prosecution, with one or more RCEs (Request for Continued Examination), this means a cost around $40,000 total could be saved if a patentability search shows the invention is not new. 
2. A patentability search on an application that is filed can save at least one office action. This alone would more than cover the extra cost of the patentability search. If the search shows the broadest claims are clearly known, the claims can be narrowly targeted to focus on a clear distinction. This eliminates the need to do an amendment to do this, and then have to subsequently argue against a rejection of that narrower claim. In addition, if you avoid the broad claim that is clearly shown by prior art easily findable, you avoid putting the examiner in the mindset that you haven’t done your homework and don’t know what you are talking about. Putting the examiner in that mindset would make the narrower claim more difficult to get than if you led with it in the first place.
3. A patentability search improves the quality of the patent. It is surprising how often inventors are unaware of prior art – they are aware of products, but not earlier patent publications and many articles. By making them aware of the prior art, they will often provide additional disclosure you might not have obtained otherwise, to distinguish the prior art. This provides support for claims where otherwise you might end up with an overly narrow patent, or have to abandon it because all the features described in the application are described are in the prior art. I’ve seen inventors mention distinctions from prior art in an office action, but the distinctions weren’t in the application.
4. A patentability search provides art for an IDS (Information Disclosure Statement). If you get an examiner that doesn’t do a good job, the issued patent may be very susceptible to attack later. By citing the prior art from the search, you make it more difficult to attack the patent later based on that prior art.  
5. A patentability search does double duty as a mini-FTO. While a Freedom to Operate review is more expensive, and focuses on the claims rather than the disclosure, a patentability search will often turn up much of what would be revealed in an FTO search. Any problem claims can then be dealt with early on, by design around or a non-infringement/invalidity opinion. This will also show a level of diligence that will be helpful for defending against willfulness charges for different patents, perhaps on different products, that weren’t found. In certain instances, this may show that a full FTO is warranted.

Whether you do it in-house or ask outside counsel to do it, and whether it is quick and dirty or more extensive, a patentability search is usually more than worth the extra cost.