A while back at a previous job I ran into an interesting problem. The marketing team was in disagreement about the overall effect of outbound links to the business. On the one hand is the basic notion that an outbound link is a way to send someone away from your site - seemingly not business-friendly. On the other hand attempting to wall in your traffic could be viewed as not very customer-friendly.
In our case, we were acting like brokers, offering quality and price comparison for the same service offered by many different companies. I wanted to put outbound links on our company detail pages to those companies, thinking it would do several things:
- Help consumers learn more about the company - a customer-friendly aspect of the site that would leave the customer feeling like they had a good experience
- Create a measure of trust between our company and the consumers - linking directly to their website would demonstrate a measure of confidence in our product pricing
- Establish a visible, online relationship between us and these companies - make Google and other search engines aware of these relationships and reinforce what our site is all about
The counter argument was that we were essentially sending our customers away to these companies where they might buy direct. We would lose the brokerage. We tossed around the idea of working out credit arrangements with our partners for traffic referrals, but we had so many companies each contract might vary and the solution seemed expensive to both hammer out and maintain.
There was also concern about a loss of paid traffic. Customers acquired through paid marketing channels might leave the site, and this was viewed as a real loss. I always found this position interesting. The upfront cost of acquisition through this channel somehow diminished the value of the "unpaid" channel. The implication, whether intentional or not, was that it was OK to lose organic search traffic but not paid search traffic.
To solve the knot, we conducted A/B/C testing on our navigation and links. Test A allowed the user to go nowhere but through our funnel. Test B allowed users to navigate throughout our site. Test C had outbound links to our partners. And without even measuring the effect on organic search ranking we found that users who were allowed to freely navigate not only converted better but bought other products on our site. Test A was in fact failing to advertise the rest of our site. Test B and C fared equally, suggesting no danger in outbound linking.
If we stop fearing the dangers of outbound links we are free to look at other benefits like how it affects SEO. SEOMOZ did a good whiteboard on this last July. And Matt Cutts talks about how PageRank flows in this blog post of his from last June.
Aside from these points of view, if you think about how a search engine spider might understand what a site is all about, imagine your site as a person with reputation. What you recommend says a great deal about your tastes. The web never makes sense in the context of websites in isolation. The nature of the web is relationships. And the best sites make great relationship recommendations to both the users and the search engines.