A proposal scored on per-seat list price rewards the vendor best at hiding cost below the headline. The lowest sticker can carry the highest three-year total once implementation, uplift, AI consumption and overage are added, and the only way to see that is a normalized total cost of ownership model that every bid is forced into. This article sets out how to build it and weight it. For the full sourcing sequence, see the ITSM procurement playbook from requirements to signature.
Why is list price the wrong basis for scoring?
Because list price is the one number the vendor controls and discounts most visibly, precisely because the margin sits elsewhere. The lines that decide the real bill are below the headline, and they differ enough between vendors that the cheapest sticker is routinely not the cheapest deal.
| Cost line | Visible in list price? | Why it moves the total |
|---|---|---|
| Per-seat subscription | Yes | The headline, and the most discounted |
| Annual uplift | No | Compounds across the term, can erase a year-one discount |
| Implementation and migration | Rarely | One-time but large, varies widely by vendor |
| AI and consumption charges | No | Metered and variable, often uncapped |
| Premium add-ons | Partly | Tier or pack fees attached to seats |
| Overage | No | Triggered above committed volume, priced at a premium |
How do you build the TCO model?
Build one model, normalize every bid into it, and let no proposal answer in its own format. The normalization is the point: same term, same seat count, same currency, same assumptions.
- Set the envelope. Fix the term (three years is the common basis), the seat counts by type, and the currency. Every bid is scored against this, not its own.
- Capture subscription by tier and seat. Itemize fulfiller, occasional and requester seats per tier, so reclassification savings are visible across vendors.
- Compound the uplift. Apply each vendor's annual increase across the full term. A low year-one rate with a high uplift is exposed here.
- Add the one-time lines. Implementation, migration and integration, stated as fixed figures, not ranges.
- Model AI and consumption. Include metered AI, consumption units and overage at a defined usage assumption, with the rate and cap locked.
- Total and normalize. Produce one three-year total per vendor on identical assumptions, with every assumption stated.
How should cost be weighted against features?
Keep cost and function on separate scorecards, and weight cost on the multi-year total rather than year one, so a vendor cannot buy the deal with a discounted first year while carrying the higher lifetime cost. How heavily cost should weigh depends on how commoditized the requirement is, which varies by organization. The way to surface the hidden cost lines before scoring is in how to structure ITSM demos to expose hidden costs, and the engagement runs through our contract negotiation service on fixed fee or gainshare.
The gated CIO and CFO Guide to ITSM Cost includes the TCO model structure and the assumption checklist behind this scoring method.
How do vendors game a TCO model, and how do you block it?
A vendor that knows it will be scored on total cost will try to move cost out of the comparable lines and into ones the model misses. Four tactics recur, and each has a structural block.
| Tactic | How it inflates the real cost | How to block it |
|---|---|---|
| Year-one discount, steep uplift | Wins the headline, recovers margin across the term | Compound the uplift in the model, score the multi-year total |
| Free or discounted implementation | Recovered through inflated subscription or services | Itemize implementation and subscription separately |
| AI rate left as a footnote | Uncapped consumption billed after signature | Lock the AI rate, cap and overage into the scored model |
| Scope thinning | Lower price by quoting a lighter feature set | Mandate the feature floor every bid must clear |
The common defense is to give the vendor no room to answer outside your structure. A mandated response format, a stated feature floor, and a rule that any open assumption is scored at the worst credible case together remove the gaming surface. A vendor that will not commit a rate in writing has told you what that rate will become, so score it accordingly until the commitment arrives.
What to do next
Build the cost model before you issue the RFP, not after the bids land, so every vendor responds into your structure rather than their own. Lock the term, seat counts and currency, mandate a fixed response format, and require each assumption to be stated. Then score on the three-year total, with any open line costed at the worst credible case until the vendor commits.
Frequently asked questions
- Why is list price the wrong basis for scoring proposals?
- Because list price is only the visible layer. The real bill includes implementation, annual uplift compounding across the term, AI consumption, premium add-ons and overage charges, and these vary widely between vendors. A proposal with the lowest per-seat list can carry the highest three-year total once those lines are added.
- What should an ITSM TCO model include?
- At minimum: subscription by tier and seat type over the full term, the annual uplift compounded, implementation and migration, premium add-ons, AI or consumption charges with overage, and any platform or integration fees. Normalize all bids to the same term, seat count and currency so the comparison is like for like.
- How should the cost score be weighted against features?
- Weight cost on the multi-year total, not year one, and keep functional scoring separate so a vendor cannot win on a discounted first year while carrying a higher lifetime cost. The exact weighting depends on how commoditized the requirement is, which varies by organization.
- How do you stop vendors gaming the model?
- Mandate a fixed response format, require every assumption to be stated, and lock the uplift, AI rates and overage terms into the model rather than accepting them as footnotes. A proposal that leaves those open is scored at the worst credible case until the vendor commits.
Carry the model past award into the contract itself. The figures that won the scoring, the uplift cap, the AI rate, the overage terms and the feature floor, are only worth what the signed agreement holds them to, so map each scored assumption to a contract clause before signature. A TCO model that decided the award but does not appear in the contract is a comparison the vendor is free to walk away from once the ink dries.
Score every bid on the real number.
We build the normalized TCO model, force each proposal into it and weight on multi-year cost. Fixed fee or gainshare.
Book a contract negotiation review →