The “A” Word in Value Pricing

A once taboo topic is now out in the open for pricing AI

Photo by Aneta Pawlik on Unsplash

In high school we read The Scarlet Letter by Nathaniel Hawthorne. Set in 17th-century Puritan Boston, the central character, Hester Prynne is forced to wear a scarlet A in order to publicly shame her for committing adultery. Clearly a taboo subject back in those days.

Here in the 21st century, pricing strategists have their own scarlet A when it comes to value prices. It may not be as public or as shameful as Hester’s but it’s a topic that’s usually avoided.

What does the “A” stand for? It stands for Attribution.

What is Attribution?

Attribution is your product’s specific value claim. Take advertising for instance. There is this venerable quote: “Half the money I spend on advertising is wasted, the trouble is I don’t know which half.” That’s the essence of the dilemma.

Another example: a company buys sales training for its team and in subsequent months revenue increases by $1 million. How much of that increase is attributed to the training? All of it? Most? A little?

The revenue uptick may have been driven by new advertising, or from the release of new product enhancements, or from a recent strong showing at a major trade show, or just plain luck, or so on, and so on and so on. I have witnessed many such debates. In worst case scenarios they can devolve into a “how many angels can dance on the head of a pin?” like futility.

And that is why some pricing folks tend to sidestep the attribution topic. Nevertheless, value attribution is unavoidable if you want your B2B product to successfully compete on differentiation.

Perspectives about Attribution

Let me contrast two different points of view about attribution, using the above training example.

1. Optimists: This POV is best represented by the vendor’s sales team. They claim that the sales training was largely responsible for the revenue increase because it improved the participants’ skill and confidence. The post training feedback scores confirmed it. Plus, this experience has been replicated with other clients.

2. Skeptics: This is best represented by the buyer’s procurement department. They argue that the company made other investments in the sales team such as a new CRM system which streamlined the sale process. Not to mention a general industry-wide upswing in purchasing. So yes, the training probably helped in closing a few deals, but the overall contribution was minimal.

Which side do you favor?

Can Attribution be Scientifically Proven?

A consulting client once challenged me with: “how would you prove the success of a project?” To which I replied with “To measure success, you get what you pay for.”

What I mean is that if you truly want proof you need to commit to defining and measuring that proof. This requires time and resources. So in order to settle the sales training debate, the company could conduct a win/loss study.

I believe that many demands for “scientific” proof of value were a bit disingenuous anyway, a bargaining ploy. Think of how much money pharmaceutical companies spend to scientifically prove the safety and efficacy of their drugs to the FDA. For those products that’s mandatory and the cost is passed on to the customer. But, do you seriously believe that companies are willing to bear the additional cost of providing that level of proof for training or a CRM? Not likely.

The New Frontier is Outcome Pricing in AI

In my nearly 20 years in the pricing space, I have seen adoption of value pricing migrate from complex manufacturing (chemicals, specialty components, etc.) to enterprise SaaS and most recently to AI products. During this journey approaches to attribution evolved.

Back in the manufacturing value model days, attribution was less controversial. Mainly because of the high capital investments in manufacturing plant and equipment, there was a lot of process measurement going on, so it was relatively straightforward (although at times complex) to estimate cost savings in time, materials and risk of chemical X versus alternative Y. Likewise for estimating the cost of waste, downtime or potential revenue increases from incremental production volume.

As the SaaS industry began to mature in the mid 2010s, there came a sudden interest in value pricing to ensure growth in an increasingly competitive space. Ironically, though digitization has unleashed much more data in the enterprise, attribution became more challenging. Compared to manufacturing, internal business processes in SaaS are relatively looser and more dependent on human resources who tend to be less standardized or predictable than machinery.

In addition, as the IT stack became more complicated, value models now had to sort out which particular piece of technology was most responsible for a given business result. This was a situation I encountered working with a marketing automation app that was integrated with other apps. Despite this complexity, SaaS companies with significant differentiation could make a compelling value case.

Everyone knows that AI technology has recently disrupted nearly all industries, especially SaaS. AI has blown-up the traditional seat-based pricing, in favor of credit-based pricing, which is a form of usage and cost-plus pricing. Many pricing experts now predict that outcome-based pricing is the future for AI applications.

For more articles on AI and pricing, see:

AI Pricing Has a Value Problem

Is Artificial Intelligence Good for Pricing?

A Possible Attribution Solution

Adoption of outcome-based pricing puts the value attribution issue front and center. It is important to understand that outcome-based pricing is closely related to value-based pricing, but they are not the same thing. Simply stated, some outcomes provide business value, meaning that their improvement increases a company’s profits, while other outcomes do not.

Take for example, an AI agent that is priced per outcome of resolved technical support tickets. The implied value is that AI agents cost less than a human customer support person. But what if, as happened to me recently, the AI agent “resolves” issues so poorly that customers churn. In this case the pricing and usage metric (resolved calls) does not align well with the value metric (customer satisfaction). So outcome yes, value no.

In cases where outcomes are truly aligned with value then I see this as an historic breakthrough for value-based pricing. Because it checks all boxes of the value pricing trinity: usage, outcomes (value) and pricing are all aligned on the exact same metric - and it is being continually measured too.

Yet even with this breakthrough, the “proof of attribution” problem still remains. Steven Forth and others have recently promised to tackle this thorny issue by applying a combination of collaborative game theory, causal modeling, and temporal difference learning frameworks. For more details on this ambitious project, see:

Steven Forth: The Value Attribution Problem and Pricing

Steven’s approach may be a bit too abstract for some, but I applaud his effort and look forward to seeing what comes out of it. If successful, he and his collaborators may one day find a way to actually provide scientific-level proof of value. Imagine that!

But Wait, There’s More!

There are practical, accessible ways to deal with value attribution. Next week I will describe my own hacks I use to make reasonable value estimations that are acceptable for customers.

Don’t let the lack of perfect information block you from estimating your value. That will cost you money!