The third alternative: productised services
First of all, what is a productised service? It's a service which you've systematised and supported by tools, automation, processes, etc, so that you've decoupled the benefit given to the client from the amount of time spent on your side. In other words, whereas in a services company the ratio of X units of time for Y units of income is relatively fixed, in a productised service, X/Y can be all over the place. Some clients will be extremely profitable, and others less so (but still worth serving - otherwise, turn them away, of course).
Accounting services are a great example of productised services. Though many accountants will charge for time above and beyond their "standard service", most of them have packaged things like "yearly accounts" or "VAT returns" into a fixed price deal. This leaves them free to optimise the delivery of those services so that they take a minimum amount of time, while still charging the client the same amount.
Even large consulting companies, like Accenture, try to productise their services. Back when I was there, Accenture was very keen to sell what they called "Managed services", where they would take over an entire function of the business and deliver it for a fixed price, enabling them to manage the costs internally and deliver the service in an efficient way without undercutting their own revenues.
Productised services don't have to, and in many cases, shouldn't, be marketed as such, or else clients may try and push down your prices if they think it doesn't take you that long to deliver (conveniently forgetting the time it's taken you to systematise the service so it can be delivered more efficiently). "But you're getting a lot of value out of our service" doesn't always work, especially not with smaller companies, so think carefully before marketing your productised service as such.
Productised services have a number of advantages. Similarly to products, they can scale much more easily than services. Once a service is properly systematised, it is easier to actually carry out, which makes recruitment much easier, since you don't need your people to be as highly skilled. You won't be able to deal with a sudden million orders, but you can ramp up capacity pretty damn quickly if you need to.
Like products, the margins can also be quite high, because most of the time-consuming parts of the service have been automated or simplified so that the human time spent is small compared to the return. Unlike products, however, the process doesn't move along without human involvement, which means you have to keep working on it - but you can structure productised services so that there is a recurring component, which provides similar benefits.
Since this is a service (and perceived as such), people naturally understand the value of it and are willing to pay real amounts of money (in the thousands or tens of thousands) which they would not be likely to throw into a product by a small company.
Because it is a service, unfortunately, you must do the sales directly - it is difficult to sell services without any salespeople. However, the advantage is that you can tailor your pricing to the type of client, and how much value your productised service brings them. If your service will provide Â£100k of value to one client and Â£1m of value to another, it stands to reason that your proposal for the second client will have a markedly higher price tag than the first (which might be based on a percentage, or some other calculation method). This is difficult to achieve with most typical startup products, since you often don't know who you're selling to until too late.
Finally, another advantage of productised services over both products and services is that they can be taken in either direction if needed. If customers start to require a lot of bespoke work, you can evolve towards a normal service model. If, on the contrary, the customer base just keeps growing, you can automate more and more of the service until it becomes self-service. This makes productised services a great way to kick off a product business, by generating these product-related revenues early on and using them to continue building out the self-service aspects of the product (in effect cannibalising your own service with your product).
Productised services are not good at all stages of a company. Google could not and should not run AdWords as a productised service, for example. At that scale, you need maximum automation. But they could have done so in the early days of AdWords. Nor are they always possible in the very early stages, before you have any idea what your market wants (but then, why are you starting a business in that industry?).
I'll address the paths to a productised service, from either a product or a service, in later articles, but hopefully in this article I've made the case for why there is a third model, which sits in between products and services, and which should be worthy of your consideration when trying to figure out how the hell you'll get your company off the ground, particularly if you're a new entrepreneur and are taking my advice to stay away from investment until you have your basic business skills figured out.
If you read this far, you should follow me on twitter here.