The software market is growing faster than companies’ willingness to accept further price rises. In July, Gartner raised its forecast for global software spending in 2026 to $1.47 trillion, 15.5 per cent more than the previous year. This growth is primarily driven by AI, the cloud and the expansion of digital infrastructure, but technology budgets are coming under increasing pressure from cost inflation, new billing models and parallel funding for AIprojects .
This is particularly evident in the SaaS market. Vertice, based on over $75 billion in managed expenditure, calculated its SaaS price inflation index at 16.4 per cent in June 2026. In April, it stood at 12.1 per cent. This is not an official index for the entire market, but rather transaction data from a single procurement platform. However, it illustrates the scale of the changes that procurement departments are facing when renewing contracts.
In this situation, negotiating SaaS solely in terms of discounts becomes rather imprecise. It is far more important to determine how much the company is paying for the service it actually uses.
A licence purchased does not necessarily mean a licence in use
Zylo analysed over 40 million SaaS licences and $75 billion in expenditure. According to the company’s data, on average 36 per cent of licences remain unused. At the same time, as much as 81 per cent of SaaS expenditure is currently controlled by business units, whilst IT directly manages only 15 per cent.
Three figures are therefore needed before renegotiation: the number of paid-for licences, the number of active users, and the actual intensity of product usage. Simply logging in once a month does not tell the whole story. For a CRM tool, the number of active sales representatives and transactions carried out may be relevant; for an analytics system, the number of queries and data volume; and for AI software, the number of tasks generated, tokens or API calls.
Added to this is the structure of the subscription packages. A company may be utilising 90 per cent of its purchased accounts, yet still be overpaying if the majority of users have the Enterprise version but are using features available on a cheaper plan.
AI is changing the economics of the contract
The biggest change today is not the price itself, but the way it is calculated. The ‘per seat’ model is increasingly being supplemented by charges for usage, credits, tokens, queries or processes executed.
This reduces budget predictability. In a Zylo survey, 78 per cent of IT leaders stated that, over the past 12 months, they had encountered unexpected costs associated with consumption-based or AI models. 61 per cent had to scale back other projects as a result.
Before signing up for the next contract period, a company should therefore be aware not only of the cost per user, but also of the cost per business unit: per transaction, per order, per customer served, per document generated or per process carried out by an AI agent. The FinOps Foundation specifically highlights unit economics as the basis for evaluating SaaS: the cost of the software must be compared with the unit that generates revenue or business value.
Without this, a cheaper rate per token or user may lead to a more expensive contract.
Price benchmarking is just the start
A company should be aware of the prices paid for comparable products and similar volumes, but a benchmark without context can easily be overestimated. Suppliers vary their discounts according to scale, industry, contract length, product scope and level of commitment. The FinOps Foundation points out that comparing SaaS is more difficult than comparing standard infrastructure services, as products deliberately differ in their functional scope.
Therefore, an alternative to the current supplier must take into account the full cost of switching: data migration, integrations, implementation, training and any potential period during which two systems are run in parallel. Only such a TCO reveals whether the threat of switching providers is economically viable.
The most valuable information is often contained in the contract
Before negotiations, it is also necessary to extract the mechanics of future costs from the contract: the renewal date, the notice period, the rules for automatic renewal, price increase limits, price thresholds, minimum commitments, fees for exceeding limits and the possibility of reducing volume.
The FinOps Foundation recommends combining contract data with current usage and demand forecasts well before renewal. This is particularly important for contracts where the number of licences cannot be reduced during the contract term, or where doing so incurs a penalty.
In 2026, therefore, negotiating power will not stem from the ability to secure a further few per cent discount. It arises earlier: when a company is able to pinpoint exactly what it is using, how much it costs, what volume will be required in a year’s time, and how much switching suppliers would actually cost.
