SaaS Network Effects Start Charging Up…
Posted by Bob Warfield on December 5, 2007
One of the things that makes Social Networks exciting is their network effect. Each new user adds more value to the network than the prior user; the value is in the connectedness of the thing, not the software. This means it gets easier the larger such networks grow because they deliver more value as they get larger. It’s easier to roll a snowball down the hill than a heavy rock up the hill. Crossing the chasm is all about the heavy rock. A startup gets a great idea that only appeals to early adopters, and it gets harder and harder to roll the rock up the hill as later stage adopters become the target audience.
Perhaps the best example of the awesome power of network effects I’ve ever come across is eBay. Once upon a time there were many auction sites, now there is essentially one because of it. The dynamics worked something like this. Each additional buyer meant more bidders on each item for sale. That meant the item was more likely to sell and at a higher price. This attracted more sellers, because they could make more money. That meant more of a selection for buyers, and it limited how high prices could go. So more buyers came to bid. That’s what I call a virtuous cycle, and eBay rode it fast and furiously to the very successful place they’re in now.
SaaS is now starting to show signs of similar network effects. Salesforce is rolling out customer data sharing, which it hopes will generate this sort of advantage. The idea is to make it easy to securely share information from your Salesforce application with other companies who are (presumably) your partners. For example, you could share sales leads. Given that Salesforce will shortly have 1 million users, you can see where the eBay analogy starts to fit in.
Larry Dignan sees this as Salesforce “creating a moat around it’s customer base” to create a competitive advantage and barrier. Given that the pricing is $100 per month per company users invite to share, I think Salesforce is couching this more as a revenue opportunity than a competitive barrier. It’s hard to believe that adding a few database records within their multitenant system to allow information cross pollination could be remotely that expensive. George Hu, Salesforce’s EVP of products and marketing says they’re pricing it for mass adoption, but that seems a stretch when you consider each company you share is the equivalent of a very expensive salesforce seat. Given their average customer has 20 or so seats, each partner they want to give access counts as another seat. Nevertheless, it may still be a valuable enough service to be well worth the price. I’ve been surprised before at how expensive some Salesforce offerings, such as Force, are. People tell me this is an artifact of their Oracle-based culture. Certainly Oracle has never been a low cost provider of anything!
Another example of network effects for SaaS was mentioned by Zoli in his blog. He was talking about the value of benchmarking information. This is an area that is hugely valuable, and the information is very hard to get. Whatever it is you do, imagine being able to get benchmarks on how your industry does the same thing. The most obvious is benchmarking on salaries and other compensation. We all want to track that information down when it comes time to agree on a pay rate! But almost everything could benefit from some benchmarking. Take corporate travel. It would be interesting to know how your company fares (no pun intended!) relative to others in terms of whether you are getting good deals on air fares and hotels. What about meals? What would you do if you learned your staff were spending in the 99th percentile on meals during business travel, for example?
Suppose you have a SaaS company doing some sort of analysis for your customers. You could provide them additional insights by showing them benchmarks for various metrics based on vertical industry. Concur does expense reporting and travel, for example. They could give you the comparitive travel information I mention above. Lucid Era does reporting on the sales cycle. They could tell you a lot of interesting information about how your sales cycle and pipeline compare to others. Xactly reports on sales compensation. Would you like to know relative to your vertical industry peers whether you pay your sales people more or less to close each incremental $1 of revenue?
SaaS companies are in a unique position to create benchmark information by analyzing the data in their multitenant systems. To succeed, they need the willing permission of the real owners of the information, lest they suffer a Facebook Beacon-style backlash. However, given reasonable privacy safeguards, it is often possible to secure permission. The most obvious safeguard is that benchmarks are aggregates. You’re presenting the average of a large number of datapoints, so the likelihood of discovering an individual datapoint is low. What does this have to do with network effects? The more datapoints, the better the benchmark.
I’ll leave you with a last example of SaaS network effects. This has to do with datacenters. SaaS runs software for you in the cloud. But as someone once remarked to me, there are many clouds. Each corresponds to a datacenter. There are advantages to being in the same datacenter in terms of latency and speed of data transfer. Certain kinds of applications can benefit from this, so eventually, it there will be advantages in which SaaS cloud you choose to trust your data to based on which applications live in that cloud. We’ve seen one example of this already. Joyent was able to offer free Facebook application hosting because it had dedicated high speed fiber to Facebook’s datacenter.
Stay tuned for lots more network effects in the SaaS world. The model is tailor-made for it.
Phil Wainewright picks up on my impression too: the SFDC data sharing option is extremely expensive. So much so that he calls it the “anti-viral” network, implying the cost is so high as to thwart the network effect. I think he is right that they’ve made a mistake there.