How to Forecast Cloud Spend Without a Spreadsheet Marathon

To forecast cloud spend, start from your current run-rate, apply the growth trend from recent months, layer in known commitments and seasonality, then subtract savings you are confident you will capture. You do not need a quarter-long spreadsheet exercise to get a forecast that holds. A good cloud forecast is built from a handful of inputs you already have: what you are spending now, how fast it is changing, what is contractually fixed, and what is predictably lumpy. The goal is not perfection. It is a number finance can plan around and a variance small enough to catch surprises early.

Cloud spend is hard to forecast for the same reason it is hard to control: it is variable, decentralized, and driven by engineering decisions that finance does not see in advance. The method below turns that variability into a defensible projection.

1. Start with run-rate

Run-rate is your baseline: take recent actual spend and project it forward as if nothing changes. If the last three months averaged $185,000/mo, your naive annual run-rate is $2.22M. Run-rate is not your forecast, but it is the anchor everything else adjusts from.

Use a clean, recent window. A single month can be distorted by a one-off, so average two or three months and exclude obvious anomalies. Accurate cost reporting is what makes the run-rate trustworthy.

2. Apply the trend

Spend is rarely flat. Look at the month-over-month growth rate over the last six months. If you are growing 4% per month, a $185,000 month becomes roughly $234,000 six months out, and that compounding matters. Conversely, if a big migration just landed and growth is flattening, do not project last quarter's steep climb forward.

The discipline is to use the recent trend, not the whole-year average, because cloud growth changes pace. A product launch, a new customer cohort, or a migration all bend the curve. Forecasting per team or product, rather than one aggregate line, makes the trend cleaner, which is another reason cost allocation improves forecasts.

3. Layer in commitments

Some of your spend is contractually fixed and should be treated as known, not estimated. Savings Plans and Reserved Instances lock in a baseline at a known rate. Annual SaaS contracts are fixed for the term. Reserved capacity and committed-use discounts the same.

• Committed baseline. The portion covered by Savings Plans or RIs is predictable, so forecast it at the committed rate, not the on-demand rate.
• Upcoming renewals. A $240,000/yr enterprise contract renewing in Q3 is a known step in the forecast. See renewal management.
• Commitment expirations. When a Savings Plan expires, spend reverts to on-demand unless you renew, which is a forecast jump people often miss.

4. Account for seasonality

Many businesses have predictable lumps. Retail spikes in Q4. A tax product peaks in spring. Batch and analytics workloads cluster at month-end and quarter-end. If your traffic is seasonal, your cloud spend is too, and a flat trend will under or over-shoot.

Look back 12 months for recurring patterns and bake them in. If last December ran 35% above the November baseline, expect the same shape this year, adjusted for growth. Seasonality is also where forecast accuracy is won or lost, so track it as a metric.

5. Project the end of the quarter

Finance usually needs two horizons: the end of the current quarter and the next 12 months. The end-of-quarter projection is the high-stakes one because it is close enough to act on. Take month-to-date actuals, project the rest of the month at the current daily run-rate, add the remaining months at the trend, and adjust for any known events.

A useful habit is to recompute the end-of-quarter projection weekly. If the projection drifts above budget mid-quarter, you have time to act. If you only check at quarter-end, you find out when it is too late. This is the same "before the invoice" logic behind anomaly detection.

6. Factor in savings

A forecast that ignores planned optimization will overstate spend. If your team is mid-way through right-sizing that will cut $8,000/mo, or about to buy a Savings Plan that saves $14,000/mo, the forecast should reflect savings you are confident you will capture.

Be conservative here. Only subtract savings that are committed and on a date, not aspirational. A forecast that assumes 30% savings that nobody has scheduled is fiction. See the concrete levers in our guide to reducing your AWS bill and rightsizing.

Putting it together: a worked example

Start with a $185,000/mo run-rate. Apply 4% monthly growth to the variable portion. Hold the $90,000 committed baseline flat at its Savings Plan rate. Add a $20,000 step in Q3 for a SaaS renewal. Add 15% seasonality in Q4. Subtract $8,000/mo of right-sizing savings landing in month two. The result is a month-by-month forecast you can defend line by line, instead of a single guessed number.

The point of breaking it into inputs is that when actuals diverge, you can see which input was wrong: was it the growth rate, an unplanned anomaly, or a missed renewal? That is what makes the forecast improvable. Forecast accuracy itself is a key FinOps metric.

Make it continuous

A forecast is not a one-time deliverable. The best practice, straight out of FinOps, is to refresh it continuously as actuals come in, compare forecast to actual, and tighten the model each cycle. Over a few months your variance shrinks and finance starts trusting the number. This is exactly what finance teams need to plan with confidence.

Costanalyst connects your cloud and SaaS spend read-only and builds a continuous forecast from your run-rate, trend, commitments, and seasonality, then folds in the savings it has already surfaced. See budget forecasting to project the quarter without the spreadsheet marathon. It never moves money; it just helps you plan accurately.