Lock-in is several things:
The economic cause of lock-in is the switching cost that the customer must overcome to move to an alternative product or technology. Switching costs include the book value less resale price of the old technology, plus purchase price and installation for the new technology. An important intangible cost may be service disruption, for example in moving to a new computer system or network switch. To be successful, new technologies must represent a sufficient advance in performance, cost, or functionality to overcome these switching costs.
Lock-in strategies often involve complementary products, like a browser and server, programmable hardware and its software, or a printer and printer supplies. The advantage from the vendor in selling one product at a high margin derives from the switching cost the customer would suffer in replacing the complementary technology.
Example: AT&T sells switches to the RBOCs, and then charges a high price for ongoing software upgrades knowing the RBOC would have a large cost to switch to another vendor.
Example: HP sells printers and then derives profits from printer supplies. The customer will tolerate high prices because of the cost of buying a new printer.
Example: Netscape gives away browsers and derives revenue from the complementary server software. The company operating the server will pay a high price to Netscape because of the large number of browsers with complementary features already deployed. (This example is interesting because it reverses the usual strategy.)
Example: To move from LP records to CD's, the consumer electronics industry had to overcome significant switching costs for the customer -- replacing player and buying new media. Similarly, the retailers had significant switching costs in new displays, packaging, etc. The content suppliers (music companies) are likely supportive since they can sell theie content all over again. As in moving from CD's to DVD, the switching costs can be reduced by incorporating backward compatibility; however, vendors may have difficulty obtaining the intellectual property rights.
Major points:
This latter point relates to the issue of industrial organization. There are many examples of both successful and unsuccessful strategies founded on this point:
The following simple model serves to illustrate the nature of lock-in, and in particular the incentive to the supplier to entice a new customer with incentives in order to lock them in and charge a higher price.
One unit purchased per period
C == unit cost of production (in competitive market and the absence of lock-in would also be the equilibrium price)
S = one-time switching cost for the customer
q = 1/(1+r) < 1 is the discount rate, r the assumed time value of money per period
Note: A dollar today can be invested at rate r, yielding (1+r) dollars the next period. Conversely, a dollar one period in the future is only worth 1/(1+r) = q today, a dollar two periods in the future is worth q^2 today, etc.
I = introductory price charged to new customer, probably lower than cost to lock them in
P = price charged old customer, probably higher than cost to derive profit
V = asset value of having a customer (present value of their future revenues)
Question: Given C and S, what should I and P be, assuming that the supplier and customer are just at the point of indifference to switching?
Present asset value of an existing customer:
V = (P-C)(1 + q + q^2 + ....) = (P-C)/(1-q)
From the supplier's perspective, if it attracts a new customer this period, their asset value the next period is V and their asset value today is qV, but the supplier has to pay (C-I) to attract them. The supplier is just indiferent to attracting the new customer if:
C-I = qV
From a customer's perspective, the recurring cost of staying with the current supplier is:
P(1 + q + q^2 + ...) = P/(1-q)
while the cost of moving to a new supplier is the switching cost plus the recurring cost, or:
(S + I) + qP/(1-q)
Thus, the customer is just indifferent to switching if these two costs are equal:
P = S + I
Note: If the switching cost is greater than the price, the customer would actually demand to be paid to switch, i.e. would not be attracted by a lower introductory price!
Solving these simultaneous equations:
I = C - qS
Note: introductory price must be lower than cost to attract a new customer
P = C + S(1-q)
Note: the markup relative to cost is S(1-q)
V = S
Note: the present asset value of a customer is precisely their switching cost, regardless of the time value of money!
In general this model is flawed for high technology, since it assumes an infinite stream of revenues, but in fact the current technology will usually be replaced in a relatively short time. This corresponds to a small value of q, like 0.3.