The Discount That Expires Friday
An observability vendor offers a three-year contract at $180,000 per year list, discounted 38% if signed by Friday. Current log volume is 4 TB per month and growing about 12% per month. Egress is billed at $0.09/GB on top of the subscription, and the discount applies only to the subscription line. The vendor's quote assumes flat volume for three years and includes no burst allowance. A self-hosted alternative would cost roughly $95,000 per year in infrastructure plus 1.5 full-time engineers, whose loaded cost is $160,000 each. The CFO has capped the observability budget at $250,000 per year and will not approve anything above that without a board paper. The team has no historical egress figures because the current tool does not meter it.
1. Is the discounted contract a good deal? Show the arithmetic that drives the decision.
Read the worked answer
Start with the number the vendor wants you to compare against, then correct it.
The headline: $180,000 × 0.62 = $111,600 per year subscription. That looks comfortably inside a $250,000 cap and reads as a 38% win.
The corrections:
- Volume. 4 TB/month growing 12% monthly compounds fast: after 12 months, roughly 4 × 1.12^12 ≈ 15.6 TB/month; after 24 months, ~61 TB/month. The quote assumes flat volume, so the subscription will be repriced or the volume capped - either way the flat assumption is not a benefit, it is an unpriced liability.
- Egress. At 4 TB/month, egress is about 4,096 GB × $0.09 ≈ $369/month, trivial. At 61 TB/month it is roughly $5,600/month, ~$67,000/year - and it sits outside the discount. This is the line that decides the deal, and nobody has the historical figures to bound it.
- Self-hosted total cost. $95,000 infrastructure + 1.5 × $160,000 = $240,000 plus $95,000 = $335,000/year, and that is before on-call and the opportunity cost of pulling engineers off delivery. It is not the cheap option; it is the option that hides its cost in headcount.
Conclusion: the discounted contract is very likely the right direction - $111,600 versus $335,000 is not a close comparison - but it cannot be approved on Friday's numbers, because the three-year egress exposure is unbounded and the growth curve makes it material, not marginal. The 38% is real; the $111,600 is not the three-year cost.
2. What do you do about the Friday deadline? Be specific about what you ask for and what you will sign.
Read the worked answer
Treat the deadline as a negotiating position until proven otherwise, and never sign an unbounded three-year commitment to protect a discount.
Ask for, in this order:
- An egress cap or an all-in ceiling. A contractual maximum on egress charges per year, or a committed blended rate that includes egress. If the vendor believes volume will stay flat, they should be willing to cap it.
- A volume-growth schedule. Replace the flat-volume assumption with a stepped schedule (e.g. 4 TB, 12 TB, 30 TB) so both sides price the same future. A quote that assumes flat growth at 12% monthly is not a forecast, it is a placeholder.
- Metering from day one. Require egress reporting in the first billing cycle so the next renewal is negotiated with data rather than guesses. Offer this as the reason to extend the deadline: you cannot responsibly commit three years without a single month of measured egress.
- A shorter initial term. A one-year term at the discounted rate converts an unbounded exposure into a bounded one. If the vendor refuses any term shorter than three years, that refusal tells you how confident they are in the flat-volume assumption.
What to sign by Friday: a term sheet or letter of intent preserving the 38% conditional on the above, if the vendor will take it. What not to sign: the three-year contract as quoted, because the only thing the deadline buys is the removal of your ability to test the assumptions.
Escalation: if the vendor will not move - take the board paper. A $250,000 cap and an unbounded egress line cannot both be satisfied, and that is a decision for the sponsor, not for you at 5pm on Friday.
3. The CFO asks why you did not simply take the cheaper self-hosted option. How do you answer, and what would make you reverse?
Read the worked answer
Say plainly that self-hosted is not cheaper once staffing is priced honestly.
The comparison on equal terms:
- Self-hosted: $95,000 infrastructure + 1.5 FTE × $160,000 = $335,000/year, plus on-call rotation, upgrade work, and the delivery work those 1.5 engineers are not doing.
- Managed (subscription plus realistic egress): roughly $112,000 to $180,000/year depending on volume, with the growth risk being the open question rather than the base cost.
The gap is large enough that the burden of proof sits with the self-hosted case, not the managed one. Presenting $95,000 as the self-hosted price is the error that makes the comparison look close; headcount is a real cost even when it does not appear in the vendor's line item.
What would reverse the conclusion:
- Egress grows beyond roughly $140,000/year - at that point managed total cost approaches self-hosted and the variable becomes who carries the risk.
- The 1.5 engineers cannot be hired or backfilled, in which case the self-hosted cost is not $335,000 but an unstaffed system - strictly worse than either option.
- Data residency or retention rules prohibit the managed service, which removes the choice entirely.
- Volumes flatten. At flat 4 TB the managed option stays cheap and the case for self-hosting disappears completely.
The judgement: cost comparisons fail most often by counting licences and forgetting people. Answer with the fully loaded number and name the single variable - egress growth - that could flip it, so the CFO is deciding on the real uncertainty rather than on the sticker price.