We’ve decided to switch to tracking usage based on the number of voters in the last 30 days. The advantages of this are:

• It’s more clear. Anyone who votes on your forum applies to your count.
• It’s doesn’t penalize you for users who haven’t returned in a while. You’re not penalized for a voting spike that occurred a few months ago.

Why this is a superior strategic move

They switched from charging for the potential to realize value to charging for the realized value. Buyers now only pay for the value delivered and no longer for the right to use the product and extract some value out of it. The uncertainty for buyers is much reduced. The company also demonstrates its strong confidence in the product. Their product is so good to induce users to give feedback that they are using only the voters as their metric. In effect, what they did is switch from a classic pricing scheme to an activity-based one where the metric is the voters.

The end goal of an activity-based pricing structure is always to come as close as possible to a value-based one. The choice of the metric(s) is instrumental. In this case, the voters seems to be a good one, but others could have been used too (like metrics based on feedback not voters).

Room for improvement

While the enhancement is significant, there’s still room for improvement.

Adopting a threshold structure, as opposed to a continuous one based on utility-pricing (you pay what you use without thresholds effects), puts the burden on the customers to figure out and predict their level of activity. True, they can upgrade and downgrade when they want, thereby avoiding harmful locked-in dynamics. Yet, clients need to monitor their level of activity in order to avoid any surprise. They are also less likely to sign-up: they are asked to make a tough and uncertain decision where they risk to overpay. Behavioral economics has amply demonstrated that faced with such a decision, an attractive choice is to just delay it.

Disjunction effects for example are at play here. If clients don’t know which plan they might need, a significant part of them will just wait for a hypothetical prediction that will never come, and some will end up not using the service or signing up with competitors.

Continuous activity-based pricing can still be regular

The main reason to avoid continuous pricing schemes is that, supposedly, the clients want certainty. In fact, clients do want to avoid uncertainty, but continuous pricing doesn’t have to burden them with uncertainty. One trick is to charge a fixed amount but “bill” on actual variable use.

So, an ideal pricing scheme for web services like Uservoice should satisfy the following constraints:

• As close to value-based pricing as possible: needs to play on the metrics here to find the correct combination
• Utility-pricing: avoid threshold effects where users have to predict how your product will behave in their specific situation. They don’t know your product well and it is a difficult choice to make. They will delay it, then overshoot and pay too much or undershoot and be disrupted by having to upgrade at a short notice.
• Reduce uncertainty for clients as much as possible. This means capping the maximum charge for very big clients.

Here’s how you could implement this:

Ask clients to estimate their needs (providing estimates of voters in increments of $10), and make it a regular monthly payment. Each month, charge them the full regular amount, but credit them back in their Uservoice account the difference between what is charged and what is actually used. If activity ramps ups, then use the accumulated credits before asking them to upgrade. If the activity is stable, then when accumulated credits reach the full amount of the monthly payment, use the credits to pay it and do not charge them that month. Even if they do like regularity, no one will complain…

If clients are above their limit, tell them the difference will be credited as negative credits, and that they will ask to pay those credits, the deadline being when they reach the amount of the regular payment. Advise them they should increase their monthly charge so that payments are smoothed out, but accept one-time payments.

Such an implementation is possible with the major payment gateway providers and doesn’t increase the total costs of processing.

Continuous pricing doesn’t have to be uncertain.

Gist

When people do not know with certainty if an event A will occur or not, and if they have to make a decision based on event A occurring, then they may postpone their decision until they know with certainty if A has occurred, even if they would have made the same decision whether or not A had occurred. As clearly put:

“We call the above pattern a disjunction effect. A disjunction effect occurs when people prefer x over y when they know that event A obtains, and they also prefer x over y when they know that A does not obtain, but they prefer y over x when it is unknown whether A obtains or not.”¹

Disjunction effects are at play in many decision-making situations and can influence how long peoples are taking to decide, and even the decisions themselves.

Origins

To understand intuitively what a disjunction effect is, let’s look at the following example:

“For another example of nonconsequential reasoning, imagine that you have agreed to bet on a toss of a coin in which you have equal chance to win $200 or to lose $100. Suppose that the coin has been tossed, but that you do nit know whether you have won or lost. Would you now want to play this gamble for a second time? Alternatively, how would you feel about accepting the second gamble if you knew that you lost $100 on the first gamble? And finally, would you play the second gamble having discovered that you won $200 on the first gamble? We have shown that, contrary to STP, a majority of respondents accepted the second gamble both after having won as well as having lost the first, but a majority rejected the second gamble when the outcome of the first was not known. This pattern—accept when win, accept when lose, but reject when do not know—was the single most frequent pattern of preferences exhibited by our subjects.”¹

Another example studied by Shafir and Tversky is a modified version of the prisoner’s dilemma where the subject is playing one-round-only PD with an unknown opponent. The twist is that he is sometimes given the decision of the opponent (whether (s)he competes or cooperates). What is observed is that, when the choice of the opponent is known, the subject chooses “compete” the vast majority of the time, regardless of whether his opponent has chosen “compete” or “cooperate”.

When he does not know, however, the proportion of subjects choosing to “cooperate” is rising significantly. Shafir and Tversky attribute this to an effort by the subject to make his best to bring a mutually beneficial situation. When the situation is played out however, this argument is no longer valid and the choice is made purely from an individual perspective. This is the same mechanism that makes us vote, even though we know our impact won’t change the result, but make us totally uninterested in voting as soon as the outcome is known.

Another explanation put forward is the following:

“Alternatively, people may consider all the relevant outcomes but, due to the presence of uncertainty, may not see their own preferences very clearly.”¹

At Stake

Strategy

Strategists must take care of Disjunction Effects and not delaying decisions just in the face of uncertainty around what is considered a key event. Often times, even though the event is indeed key, the decision taken is the same even after different outcomes. They would in effect delay a decision in order to wait for an event to be realized or not, even though this does not affect their final decision. Put it another way:

“One result of nonconsequential decision making is that people will sometimes seek information that has no impact on their decision.”

Subjects are in effect willing to pay for information that is not going to change their choice, but merely going to clarify their reasons for choosing.

Marketing

If you put your clients in front of a choice where the uncertainty on the outcome of an event is leading to a disjunctive choice, you risk delaying your client’s choice, and thus risk losing him or her entirely if conditions change during this delay. Price structures in particular can lead to such effects.

Corporate

Some managers can fall prey to this effect and choose to delay decisions before what is considered a decisive event, even though external conditions and fixed constraints mean that heir choice will be the same regardless of the outcome of the decisive event.

Actionable ?

Tversky and Shafir show that “once people are made aware of their preferences given each possible outcome, the sure-thing principle is no longer violated.”² Thus explicit analysis and description of the different possible scenarios along with the different course of actions decided in each case would tend to, at least, reduce the risk to fall into this trap.

More details

Disjunction effects go against the sure-thing principle (STP). “The sure-thing principle says that if we prefer x to y given any possible state of the world, then we should prefer x to y even when the exact state of the world is not known. […] STP is an important implication of the consequentialist view. It captures a fundamental intuition of what it means for a decision to be determined by the anticipated consequences. It is a cornerstone of Expected Utility Theory.”¹

References

1. (1992) Shafir, E., & Tversky, A. Thinking through uncertainty: Nonconsequential reasoning and choice. Cognitive Psychology, 24, 4, 449–474.
2. (1992) Tversky, A., & Shafir, E. The disjunction effect in choice under uncertainty. Psychological Science, 3, 5, 305– 309.