The Consumption Forecast That Protects You

Why is the consumption forecast the whole negotiation?

The consumption forecast is the whole negotiation in a usage based deal because the committed amount, not the unit rate, determines what you actually pay. Platforms such as Snowflake, which prices in credits, and Databricks, which prices in DBUs, sell capacity commitments where you agree to consume a certain volume over the term in exchange for a discounted rate. The discount looks attractive, but it is only a saving if you consume what you committed to. Commit too high and you prepay for capacity that sits idle, which is the most common and most expensive mistake in consumption pricing. Commit too low without protection and your overage runs at undiscounted on demand rates that can erase the saving. The number that decides which outcome you get is the forecast, and the buyer who lets the vendor supply that forecast has handed over the most important term in the contract.

This is a specific case of bringing your own data to every negotiation. The general discipline is set out in the SaaS Negotiation Guide, which shows why evidence beats vendor assumptions at every stage of a deal.

What does a forecast that protects you actually contain?

A forecast that protects you contains your real historical consumption, an honest growth assumption, your seasonality, and a clear separation of committed baseline from variable upside. It starts from measured usage rather than a sales projection, because the vendor's growth curve is built to maximise your commitment, not to match your workload. It accounts for the lumpy reality of data platforms, where a quarter end batch, a migration, or a one off analytics project can spike consumption far above the steady state. And it draws a deliberate line between the consumption you are confident you will reach, which becomes your committed floor, and the consumption that might happen, which you handle through flexible terms rather than a higher commitment. The goal is a number you would still be comfortable with if next year disappoints, because the committed floor is the part you are obligated to pay.

How do you build the forecast from real workload data?

You build the forecast by pulling your own usage telemetry, normalising it for one off events, and projecting a conservative baseline you can defend. The table below shows the inputs that turn a vendor guess into a buyer grade forecast.

For the underlying meter on each platform, read Snowflake credits and the consumption model, and for the contract limits that cap your exposure, see usage ceilings and consumption caps.

How do you protect a wrong forecast with contract terms?

You protect a wrong forecast with rollover, burndown, true up bands, and usage ceilings, so that an estimate that misses does not become a year of waste or a punitive overage bill. Rollover lets unused committed capacity carry into the next period rather than expiring, which softens the cost of committing slightly high. Burndown terms let you draw down a prepaid commitment flexibly across the term rather than on a rigid annual schedule. A true up band lets you adjust the commitment within a defined range without renegotiating the whole deal. And a usage ceiling or on demand rate cap limits what overage costs if consumption runs hot, so you are not exposed to undiscounted rates on every credit above the commitment. Together these clauses turn the forecast from a single risky bet into a managed range, which is exactly what a buyer wants when the future is uncertain.

How do you negotiate the commitment itself?

You negotiate the commitment by committing to a defensible floor, trading term length for rate protection, and refusing to let the vendor's growth story inflate the number. Bring your forecast and anchor the conversation on the consumption you are confident you will reach, then let the vendor earn any larger commitment with a deeper discount and the flexible terms above. Where the vendor pushes a higher commit for a better rate, model the breakeven honestly: a lower rate on capacity you never use is more expensive than a higher rate on capacity you do. Time the deal to the vendor's quarter or fiscal year end, when the pressure to close improves both the rate and the terms. And lock the discounted rate for the full term, so a strong forecast this year is not repriced against you next year.

A worked example of a protective forecast

Consider an indicative example. A retail analytics team is offered a large Snowflake capacity commitment at an attractive credit rate, sized on the vendor's projection of rapid growth. Rather than accept the projection, the buyer pulls a full year of credit usage, strips out a one time migration spike, forecasts by season, and finds its defensible baseline is well below the proposed commit. It commits to that floor, negotiates rollover so any underuse carries forward, secures a burndown schedule and a true up band, caps the on demand overage rate, and locks the discounted credit rate for the term. When growth comes in slower than the vendor predicted, the buyer is protected rather than stranded on prepaid capacity. These figures are indicative, but the structure is reliable, and a forecast led commitment lands the buyer inside the 10 to 30 percent savings disciplined negotiation typically produces, by published market estimates.

What to do next

Before you sign any usage based commitment, build the forecast from your own data, commit to a floor you can defend, and protect it with rollover and true up terms. The renewal mechanics are covered on the SaaS renewal negotiation service, and the full buyer method runs through the SaaS Negotiation Guide.

Last reviewed April 2026