Page 2 of 5
When should you outsource?
Outsourcing is a business decision about the relative benefits of outsourcing versus in-house. It's not a black-and-white issue, but exists in that wonderful grey land of "It depends." Whatever the case, you should never outsource anything you do not fully understand, particularly when it is a significant part of your expenses. If you do, you will never understand how to manage the outsource relationship. Outsourcing is not a replacement for intelligence. You are the scientist, the creator, the lab genius. Outsourcing is the staff you have to do your research, and it's funded by the research itself. If you are in charge of SEM without truly understanding it, then you better hire someone with a visceral knowledge and you better be able to understand and learn from them. They are not a replacement for you -- otherwise you're just outsourcing to someone internal rather than external.
The type of outsourcing you engage in also depends on what type of company you are.
Large clients who spend millions of dollars a month across multiple domains often completely outsource some verticals. In other cases they lease technology and occasionally use services a-la-carte. They are smart enough to know when it makes sense to outsource. Even their fully outsourced business units are sufficiently staffed to communicate heavily with the outsourced team. They take advantage of the synergies among media types, the benefits of landing page testing, and the general-offer testing that results in higher conversions.
Outsourcing is intrinsically tied to staffing. Case in point: A travel-related website was dramatically increasing its paid search spend to drive inquiries to its advertisers. They had very focused needs and had been testing a system that had the same spend, but were seeing nothing particularly special in results. They were habitual tinkerers and thought they could improve the paid/natural inquiries.
They were introduced to Marin Software, a do-it-yourself version where your in-house people use the platform. But Marin also brought relationships and expertise that smoothed the road into the usual search suspects.
Coincidently, the travel company experienced a cash flow problem and cut its budget pretty deeply for a while. As a necessity, the company started looking closer at the traffic it had and discovered that it could radically increase natural search functions with better optimization. As a result, they are spinning up the paid search campaign again, and with more attention to detail and talented technologists the company will get 100 percent of the traffic it needs for the model to work, at 25 percent of the cost. However, that comes at the expense of staffing internally.
Importing technology into an enterprise and having the staff to utilize it is "in-sourcing." You get to control the technology more discretely and make more rapid changes, but at a cost -- internal employees. This is still infinitely better than trying to build a system from scratch.
To accurately predict your savings, your SEM cost should always calculate the additional internal resources to run that technology. Unfortunately, most companies view employee costs as a "soft cost" that is not calculated into this figure. When you outsource something critical, you may find that the 10 to 15 percent (approximate outsource fees) actually costs more than you ever thought. Before you outsource, hire a consultant to guide you in this effort. It will likely be the best money you ever spend.
Why don't you build your own system?
Companies that specialize in SEM spend money -- a lot of money -- on technology that they can distribute over a wide variety of clients. Unless you are an eBay or an Amazon, both of which have a slew of technologists with a fundamental understanding of online, it's a fools' errand.
Companies build so little of their other key software (ERP, CRM, accounting, ecommerce, etc.) and instead customize the data flow in and out of these applications. Why they think APIs into the search engines require that they build the whole campaign management suite is beyond me. Most likely you'll have what you need for one millisecond, and then your technology will slowly fade into techno-archaic land. Or worse, you create a system that requires so much ongoing upkeep and modification that detaching it from your internal process becomes next to impossible. I've yet to see an in-house technology solution that comes close to even the least effective, slowest, and least exciting competition.
Often it happens because someone runs a test with a limited number of keywords and generates a profit. The mistake they make is then scaling that test based on estimates of what they could make if they built their system internally.
I have watched countless agencies and clients go down the same road. Internal arrogance makes them say, "No one knows our business like us!" That statement is true, but often it turns into a land grabbing war. SEM is so profitable, with such minimal effort compared to other marketing programs, that the person handling SEM gets lionized. Even a highly trained monkey can make a select group of keywords profitable, but it's not about that. It's about scalability and going from a handful of keywords to thousands.