Every Company is Now an AI Company
Companies of all stripes and sizes are trying to assess options for adopting AI – not only as internal productivity tools, but also for embedding AI-based capabilities into their products and services. This is both a defensive and an offensive business strategy. On the defensive side, it helps companies defend their installed base, market share and pricing power against encroachment by innovative AI start-ups. As an offensive strategy, it is a means of penetrating new market segments and pursuing opportunities to create value with AI-based offerings. No matter how established and dominant organizations might feel about their current market position, AI is forcing everyone to consider scenarios which, until now, have been viewed as highly improbable. No company will be able to skip this critical assessment of AI impact, thus forcing every company to become an AI company.
With stakeholder pressure to demonstrate not only AI initiatives and capabilities but tangible financial outcomes, AI monetization has become an imperative. The vast majority of companies are infusing AI into their products and services, regardless of vertical industry. Of those, nearly 75% are expecting these new products and services to drive significant incremental revenue growth. Converting AI solutions from a concept and a demo to a viable revenue generating commercial offering will stress-test every company’s Quote-to-Cash stack, spotlighting gaps that may exist in tools, skills, processes, data, compliance and best practices. This stress test is forcing organizations to transform their monetization capabilities.
Why Is Usage Billing Key to AI Monetization, and Why Isn’t It a Passing Trend?
Key amongst many elements required to support successful AI Monetization is the capability to support robust usage pricing, billing and revenue management. There are several secular and persistent factors driving the need for consumption-based approaches in AI monetization:
- AI products are new and untested, with pricing, packaging, ROI, and key outcomes still uncertain. Because of this, users tend to be slow to adopt such solutions at scale. Usage-based pricing approaches allow companies to align use and cost more closely than fixed recurring charges. Usage pricing has evolved as the dominant mechanism for AI product price discovery. Buyers of AI solutions are largely willing to give it a try for now but expanded production use will require much greater billing transparency from AI product suppliers. Longer term, customers value budget predictability above all else.
- Every AI product is usage-shaped on the cost side, even when flat-priced on the revenue side. Tokens are consumed per use, so cost is variable and per-customer regardless of how the vendor invoices. Therefore, AI margin is a measurement question before it is a pricing question. Even companies that have deliberately chosen not to price on usage will be forced to invest into consumption measurement. Every company will have to address the same question: “What is our AI margin?”.
- With adoption being a challenge for AI-based products, usage measurement is hardly an optional capability. Users are stuck in a state of inaction rooted in uncertainty. They do not want to be passive in the face of a competitive threat but also are cautious about over-investing into something that may be much cheaper and simpler in six months or even obsolete in a year. Boards and investors demand to see concrete evidence of adoption metrics to continue funding AI initiatives.
- Even companies that have made a deliberate decision not to pursue consumption-based pricing for AI products, rely on open-source, self-hosted AI models, and/or force users to utilize their own AI keys may be pushed to reassess their usage billing capability. A competitor can force their hand at any time, but launching usage pricing from standing a start can take 6-12 months.
The Need in AI for Usage Pricing is Established – Why is This a Big Deal?
With every company becoming an AI company, every company will become a usage-based pricing and billing company. This capability used to be the domain of telecom and logistics/transportation companies that employed usage billing experts and spent generously on software and services in support of that model. Prior to the latest wave of AI spurred by ChatGPT in November 2022, several VC-backed software companies were already touting benefits of usage pricing and billing, yet that motion by itself never really gained mainstream traction. In fact, our field research at the time frequently encountered start-up founders with successful usage-pricing business models, enviable growth, and profitability, yet they were unable to raise even a seed round of funding. The reason was simple – lack of expertise amongst most of the venture and angel investors who found these companies’ business metrics unfamiliar, not fitting their traditional subscription models and ARR metrics. Similarly, as the world rushes to adopt AI and monetize it – most likely through some combination of usage and value pricing – only about 5% of companies actually have resources (i.e., personnel) with direct experience implementing and managing Quote-to-Cash processes including usage- or consumption-based pricing models. This gap between the business requirement for usage billing and internal level of competency with consumption-based models has the potential to become a gating factor to the speed of AI adoption.
How will Companies Deal with Usage Pricing and Billing?
For the 95% of the organizations globally that do not have expertise with usage pricing and billing, the journey to a functional consumption business model will follow what looks like an obstacle course characterized by lack of knowledge, industry noise, misplaced priorities, under-estimated budgets, over- or under-confidence, and complacency.
It starts with a blind spot. Most organizations will initially assume usage pricing and billing capability already exists in their ERP system or in a dedicated billing solution that is already in place. So, the challenge of usage billing is often ignored or downplayed. Beyond just assembling the correct usage billing tools, critically assessing how to understand usage patterns to inform pricing is consistently lacking. AI pricing models now are often built in a way that inflates the initial price by including some or significant number of credits on a theory that by the time AI adoption really picks up, a solution will be found. These theories on AI credit allocation are typically built around the concept of an average user’s consumption, but the reality is that one super-user can torpedo the entire pricing scheme. For example, portfolio managers at some Wall Street investment firms can easily spend a few hundred thousand dollars on AI tokens per month. Most early providers of AI-based tools do not even get to that stage, as their first approach will be to ignore symptoms, to rationalize them away, or attribute over-use and under-counting to wrong causes.
Eventually, enterprises recognize usage billing, pricing, and revenue management are bona fide challenges. Many will initially assume the solution to this issue is very simple: that a usage billing solution already exists within the current monetization stack or that it can be either easily vibe-coded, developed, or procured and implemented in very little time. In other words, when it comes to consumption business models, they continue to underestimate the complexity, costs, and downside of consumption-based business models and overestimate current capabilities. That underestimation doesn’t resolve itself – it surfaces as symptoms: unbudgeted token bills, bad customer experience, billing disputes, and sales reps spending more time pacifying unhappy clients than selling. Revenue accounting and recognition start to break down and auditors begin raising red flags. Only at that point does the need to act become undeniable, and product, finance, sales, and IT are belatedly forced to treat usage pricing and billing as something worth real attention and budget. What follows is rarely clean: a period of trial and error, false starts, band-aid fixes, and normalized workarounds – sometimes with a full team changeover – before a real, engineered solution finally gets built.
What Makes Usage Billing a Complex Challenge?
An organization can decide to pursue usage pricing in a single meeting, but the capability to execute this takes 6-12 months to build. Most organizations prioritize pricing, not realizing that this is the simpler part of the overall problem. Complexity of usage billing can vary greatly. Most billing solutions today have some support for ingesting a batch file containing usage metrics, consolidating the charges and calculating a bill. Organizations whose business models require data based on a single variable from a single consistent and high-quality data source that can provide a simple batch file for upload into a billing system with low volume and no obligation to provide usage billing transparency during the billing cycle (only at the end), most billing solutions should provide a solution. For anything more sophisticated, the challenge complexity rises rapidly.
Fig. 1 below outlines the progression of usage billing complexity starting with the first five elemental stages.
Table 1 below outlines the five basic stages of Usage Billing Complexity
Spectrum PositionTechnical ParadigmCore Problem StatementArchitectural Focus1. The Batch Monolith (Simplest)Single-Source,Static Batch File“How do we accurately parse and apply a rate card to a single, structured file at the end of the month?”Simple ETL, script execution, batch scheduling2. Multiple Incompatible SourcesMulti-Source,Scheduled Batches“How do we reliably join and deduplicate varied batch data from distinct systems before the billing cycle closes?”Relational database joins, schema mapping, data staging3. Quick-TimeEvent-Driven Streams, Unified Schemas“How do we continuously ingest and process a high volume of usage events without losing data or creating billing lag?”Stream processing (e.g., Kafka), idempotency, queue management4. Real TimeSimilar to stage 3 but now in real time with session set up (on provision) and session take-down on cancellation or non-payment“How do we provide real-time billing transparency with high volumes and demanding customer base?”May require complete re-architecture of ...
