The sweet spot for NFV so far has been virtual CPE (vCPE) and the sweet spot for vCPE has been managed services. Nearly every operator out there has managed services ambitions at some level, but at least three out of four admit that they’re planning in a more hopeful-than-helpful sense. Is there a right, or best, way to address the opportunity? Yeah, there is, as you’ve no doubt guessed, and I’ve tried to assemble the guidance here.
To start with, managed service success is rooted in what could be called geo-demographics. Prospects naturally sell themselves if they’re associated in a geographic sense, sales are easier if you have rich territories, and infrastructure and support are most efficient where customers can be concentrated rather than scattered over a wide area. Most countries offer both consumer and business census information by locality, meaning zipcode or metro area at the least. You start there for the geo part.
For demography, you have to look at the various MSP value propositions, which means looking at the service customer you want to turn into an MSP customer. To start with, that’s an important point in itself because you don’t want to try to sell MSP services to somebody who isn’t connected already. The selling cycle will be too long. Ideally your prospect will have a service connection and have issues that can be addressed with a managed service.
Those issues are most likely to be present where there’s little or no network support expertise in-house, or perhaps even in the local labor pool. If you want to sell a managed service you’re selling management of a service, which means you’re competing with any in-house personnel who happen to provide that already. Since these very people are likely the ones to be assessing your offering, it can be a tough sell. Most consumers lack network tech skill, and so do most SMBs.
There are a lot more consumers than SMBs, of course. In the US there are about 6 million SMB sites and almost 30 times as many households. However, business willingness to pay is much higher because the equipment and support needed to sustain business connectivity is much more expensive. Census data that identifies the business population of a given area (I’ll use zipcode here as my example) is available, and most important that information is usually available by industrial classification (SIC, NAICS, etc.) within geography.
The reason you need the industry data is that the consumption of IT and network services varies radically by industry. Those that consume the most on network services spend about six thousand times those that consume the least, in the US market. What you’d like to find is an industry that spends a lot on network services and is well-represented by mid-sized businesses in your geography.
Another piece of industry data to look at is the rate of spending on integration services. The leading industry in this category, in the US, spends about 15% of their IT budget on integration services where the last of the pack spends less than a half-percent. If your market geography provides multi-year information you can also look at network spending and integration as a percent of IT spending growth over that period; fast growth usually puts stress on internal support.
When all else fails, look at the ratio of spending on personal computers to minicomputers and servers, meaning central IT. Where there’s a large central IT structure there’s likely to be more technical support available, where companies with a bunch of distributed PCs often don’t have nearly the level of support the centralized gang do.
NFV vendors and analysts may like the idea of targeting specific virtual functions for your vCPE, but my data suggests that there are three broad sweet spots and that further refinement is difficult and often non-productive. The best area is security, which includes virus scanning, encryption, and firewall services. Second-best is facility monitoring, meaning the collection of sensor data and activation of control processes, and last is management of IT exercised through the network connection.
The limited experience available with vCPE marketing so far suggests that the best strategy is to present these three categories all at once rather than to focus on something, for the solid reason that a single application in a single area isn’t likely to generate broad buy-in on the prospect side and for such an application a device might be a viable option. A good MSP salesperson would try to engage on at least two of the three categories, which means that pricing should favor that. Pick a target with easy adoption in two areas and run with it.
Be very aware of the nature of the managed services you’re offering relative to the connection model of the user. There are about 7.5 million business sites in the US, about 1.5 million of which are satellite sites of multi-site businesses. This group is a great source of opportunity for VPN services, but try selling a VPN to a company with one location! Also be aware that when you sell a multi-site business, you don’t sell the branches but the headquarters, so you need to check the business name online to see where the HQ is located. If it’s out of your sales target area, forget it.
Once you have your prospects and your target service, think about fulfillment. Your approach has to balance the cost of premises hosting on an agile appliance (vCPE) versus centralized hosting in one or more data centers. The vCPE approach has convincing credentials as the best entry strategy because cost scales with customer base, and it’s going to retain that advantage for even fairly large deployments if the customers are widely distributed. That’s because network connection and support costs for the tie-back of each user to a small number of data centers could quickly eradicate any economy of scale benefits.
Where this might not be true is where you have a very geographically confined customer set. A small customer geography means that a small number of hosting points would serve customers without much hairpinning of traffic patterns to reach the hosting points. This points out the infrastructure value of concentration; your sales types will tell you it also generates better reference accounts. That’s particularly true if you have a limited industry focus; no reference is perceived to be as valuable as a firm of the same type in the same area.
There is a decent chance that if you can concentrate prospects you could migrate from pure vCPE to a hybrid of vCPE and cloud hosting, or completely to the cloud. You’ll probably know whether this is even feasible based on the census data because that will give you a top-end estimate of what your prospect base could look like.
A final critical point to remember is that all my operator research shows that the critical question with new NFV services isn’t how well your implementation can do in reducing capex or offering “agility” but in how well it manages opex. There is no practical reduction in capex that can justify a large-scale deployment of NFV absent stringent operations efficiency measures. Even vCPE is vulnerable to opex issues, and vendors’ positioning of NFV consistently underplays opex impact. Opex control is particularly important at the “S” end of SMB and in the consumer market, where price tolerance is so great and scale so necessarily large that even small operations issues are insurmountable.
Opex will also be critical for operators who have goals too lofty for vCPE and managed services to attain. Funding a broader NFV and SDN base will be critical for those, and that funding cannot be secured through any realistic path other than opex efficiency. In fact, in large NFV data centers, the operating costs per tenant could easily run three times the per-tenant capex.
My models say it is possible to make money on managed services, and in fact a decent piece of change, but it’s not just a matter of signing up a few VNF vendors and running an ad. You’ll need a marketing campaign with proper geographic, demographic, and service targeting, and an implementation that can control operations impacts and save your profits. All of the technical and benefit issues can be resolved using offerings from a half-dozen vendors. The business issues are going to take leg work on your own.