A commitment would lock a year of flat reporting spend, but a redesign lands next quarter — how much do you commit, and against what?
answer
- what survives the redesign
- durable floor, not the flat total
- term risk against discount depth
- ladder the purchase, stagger expiry
- upside is a fraction, downside the rate
basics
~20 sCommit only the part of the baseline the redesign will not touch, on the shortest and most flexible instrument that still pays. Flat spend proves the amount is stable; it says nothing about the shape, and a stranded term keeps billing against a workload that no longer exists.
solid answer
~50 sFlat spend for a year proves one thing — the **amount** is stable — and the question is about the **shape**, which the redesign is about to change. So split the baseline. The part that survives any plausible version of the redesign is the durable floor, and that part can carry a deeper, narrower instrument. The part the redesign might remove or move gets a shorter or more flexible instrument, or none at all. The asymmetry decides it: the extra discount from the deeper purchase is a fraction of the rate, while the downside of stranding is the whole committed amount continuing to bill against usage nobody runs. Buying in tranches that expire at different times keeps part of the decision reversible every few months. And the cost of stranding is not only the money — a live commitment quietly argues against the redesign that made it obsolete.
go deeper
Take away the core caution: a commitment keeps charging for its whole term even if the thing it was bought for is switched off next month.
Explain why a stable monthly total is not evidence of a stable resource shape, and why the part of the baseline a redesign cannot touch is the part worth committing.
Show the sizing judgment: commit the durable floor, ladder the rest so expiry dates are staggered, and prefer a flexible instrument wherever the design is in motion.
Own the bet and its governance — who may commit how much of the estate, what the roadmap has to say before a multi-year term is signed, and how the decision is reviewed afterwards.
## What stranding is A commitment is stranded when the term is still running and the usage it was bought to match has stopped existing — the fleet was resized, the service moved onto a managed tier, an efficiency project halved the footprint, the workload moved location. The instrument does not notice. It keeps drawing its charge every hour until the term ends, matching nothing. This is the risk the buyer accepted when the discount was granted, and it is the opposite of the risk on capacity the provider may reclaim, where the provider kept the right to take the machine back. Here nobody takes anything back: that is the problem. ## Flat spend is not a stable shape A year of flat spend is a genuine argument for committing something, and it is evidence about exactly one variable. It says the **amount** did not move. It says nothing about whether the same shape will still be running in six months. A multi-tenant reporting service is a good example of the gap: the monthly total can be unchanged while the design underneath goes from a fleet sized for peak to something that scales down between report runs. So the quantity to commit is not the flat baseline. It is the **durable floor** — the part of the baseline that survives every plausible version of the change: - usage the redesign does not touch at all; - usage it touches but cannot remove, only move within the same eligible set; - the level the workload never drops below even in the redesigned form, if that can be estimated. ## Pricing the option Set out the choices rather than arguing about them. Say today's baseline is 100 units per hour and the part immune to the redesign is about 60: | Purchase | If the shape holds | If the redesign lands in month 4 | |---|---|---| | Commit nothing | No saving | No saving, nothing stranded | | Commit 60 (the durable floor) | Saving on 60 units for 12 months | Same saving; nothing stranded | | Commit 100 (today's whole baseline) | Saving on 100 units for 12 months | Saving on 60, plus 8 months paid for 40 units nobody runs | The comparison that decides it is the last row against the middle one, and the two sides are not the same currency. The upside of committing the extra 40 is **the discount** on 40 units for 12 months — a fraction of the rate. The downside is **the whole rate** on 40 units for 8 months, paid for nothing. Because the loss is measured against the full price and the gain only against the discount, the extra tranche has to be quite unlikely to strand before it is worth buying. ## Structuring the purchase 1. **Buy the durable floor first**, and only that, on the instrument whose depth you would still accept if the redesign arrived early. 2. **Ladder the rest.** Several smaller commitments starting a few months apart mean part of the baseline comes up for re-decision every quarter instead of once a year. 3. **Prefer flexibility where the shape is in question**, even at a worse rate: an instrument that follows a change of shape is worth more than a deeper one that does not, exactly when a change is being planned. 4. **Write down the assumption with the purchase** — what shape it assumes, and what would falsify it — so the review after the redesign is a comparison rather than an argument. ## After it strands: the expensive part The money is the smaller half. The larger half is what a live commitment does to the next decision. Teams keep an obsolete shape running to consume a commitment that has already been paid for, and defer the redesign that would have saved more than the commitment ever did. That is sunk cost reasoning with a monthly invoice attached to it. The discipline is to separate the two decisions. The commitment is spent; whether to keep the old architecture is decided on its own merits. Then, separately, check what the platform allows for the instrument itself — an exchange into another one, a resale, or reapplication to other usage inside the same billing grouping. Some platforms offer one or more of these and some offer none; assume nothing until it is checked. ## What is actually being asked This is a lead's question because it is a bet with no right answer, made on somebody else's roadmap. What an interviewer is listening for is whether you separate amount from shape, whether you size the bet at the part you would still defend after the change, and whether you can say out loud what the commitment would cost you if you are wrong.
- How would you put a number on the risk of committing before a redesign?Compare two quantities over the term: the extra discount the deeper tranche earns if the workload holds, and the committed amount that keeps billing against nothing if it does not, weighted by how likely the change is and how early it lands. The second is measured against the full rate, so it usually dominates.
- What do you do with a commitment that is already stranded?First stop it steering design: the money is spent, and keeping an obsolete shape alive to consume it normally costs more than it recovers. Then check what the platform allows — an exchange, a resale, or reapplication to other usage in the same billing grouping. Options differ, so verify rather than assume.
- Why ladder commitments instead of making one purchase a year?Because a ladder converts one irreversible decision into several small reversible ones. With tranches expiring every few months, part of the baseline can be re-sized against what the estate now looks like, at the cost of a slightly shallower blended rate.
It is signing a two-year lease on an office the month before the team decides to work remotely. The rent is cheaper per desk and the desks are the thing about to disappear.
saying these in an interview costs you the question
- Commits the whole flat baseline because the total has not moved
- Assumes a stranded commitment can always be returned or resold
- Treats the deepest rate as the default and the redesign as a detail
- Keeps the old architecture alive to use up money already spent
- Judges the purchase by the rate obtained rather than what was consumed
- Argues flat spend proves the resource shape is stable too