The Valuation Is Only as Reliable as Its Assumptions
A valuation can be precise and still rest on an unrealistic terminal growth rate or discount rate. Check the assumption set before trusting the output.
Who this is for
Investment analysts and researchers — Analysts reviewing an AI-generated valuation model where the output looks precise but the assumptions underneath it haven't been checked
The problem
A valuation model can produce a number to the exact cent while resting on a terminal growth rate that exceeds long-run GDP growth forever, or a discount rate that doesn't reflect the company's actual risk profile. The precision of the output has nothing to do with the soundness of the inputs — and an AI-generated valuation summary rarely surfaces which specific assumption is doing the most work.
The risk compounds because valuation math is genuinely sensitive to small assumption changes. A one-point shift in terminal growth or discount rate can move a valuation by a large percentage, and that sensitivity is exactly what a clean, confident-sounding output tends to hide.
How ConvergePanel helps
ConvergePanel checks an AI-generated valuation against a specific assumption checklist across five models — growth, margin, discount rate, terminal value, share count, and currency — and flags where an assumption looks aggressive, internally inconsistent, or unstated. Where models disagree on whether an assumption is reasonable, that's the input to stress-test before trusting the output.
How they compare
| Assumption | AI-Stated Value | Reasonableness Check | Sensitivity | Reviewer Action |
|---|---|---|---|---|
| Terminal growth rate | 4.5% in perpetuity | Exceeds typical long-run nominal GDP growth used for terminal value | A 1-point change here can move the valuation by a large percentage | Cap terminal growth at a realistic long-run rate and rerun the output |
| Discount rate (WACC) | 7%, described as 'industry standard' | Doesn't reflect this company's smaller size, leverage, or business risk | Lower discount rates mechanically inflate present value | Rebuild WACC from the company's own capital structure and risk profile |
| Share count | Basic shares only | Excludes outstanding options and RSUs from the per-share calculation | Understates share count, overstates per-share value | Recompute on a fully diluted basis before citing a per-share figure |
How it works
- 1List every assumption feeding the valuation, not just the final output
- 2Check the discount rate against the company's actual risk profile, not a generic industry average
- 3Check the terminal growth rate against realistic long-run bounds
- 4Confirm which comparable multiple was selected and why
- 5Check share count for dilution from options, RSUs, or convertible instruments
- 6Run the assumption set through ConvergePanel across five models
- 7Flag assumptions where models disagree or where the AI didn't state a specific number
Use cases
- Checking whether a DCF's terminal growth rate is realistic before trusting the output
- Verifying a discount rate reflects the company's specific risk, not a generic benchmark
- Confirming a per-share valuation accounts for dilution
- Auditing an AI-generated valuation summary before it supports a position or recommendation
Twelve inputs worth checking before trusting the output
- Revenue growth — is the assumed trajectory grounded in a specific driver, not just extrapolation
- Margin assumptions — do they expand faster than the business has demonstrated it can
- Discount rate — does it reflect this company's actual risk, not an industry average
- Terminal growth — is it bounded by realistic long-run economic growth
- Multiple selection — is the chosen multiple justified against the comparable set
- Capital intensity — does the model account for the capex the growth actually requires
- Dilution — are options, RSUs, and convertibles reflected in share count
- Debt — is the capital structure assumption consistent with actual filings
- Share count — basic or fully diluted, and is it labeled
- Currency — are cash flows and comparables converted on a consistent basis
- Inflation — are nominal and real figures being mixed anywhere in the model
- Comparable-company selection — see the comparable-company-mismatch check for this specifically
Precision is not the same as soundness
A valuation output carrying several decimal places creates an impression of rigor that has nothing to do with whether the underlying assumptions are defensible. The math is only ever as good as the inputs — and valuation math in particular is sensitive enough that a small, unstated assumption can move the conclusion more than any modeling error would.
Checking the assumption set isn't about rebuilding the model — it's about identifying which two or three inputs are doing most of the work, and confirming those specific numbers before the output is used for anything.
Frequently asked questions
Why does the terminal growth rate matter so much in a valuation?
Because it compounds forever in the model's math, a small difference in this single assumption can move the valuation more than almost any other input — which is exactly why it deserves specific scrutiny rather than being accepted at face value.
Is a lower discount rate always a red flag?
Not automatically, but it should be justified against the company's specific risk profile — size, leverage, and business volatility — rather than borrowed from a generic industry figure that doesn't reflect this particular company.
How much does share count really affect a per-share valuation?
It can matter significantly for companies with substantial option or RSU programs — using basic shares instead of a fully diluted count can meaningfully overstate the per-share value.
Can several AI models validate the same flawed assumption?
Yes, particularly for assumptions that mirror commonly-repeated industry framing. That's why the check has to test specific numbers against reasonableness bounds, not just ask whether models find the assumption plausible-sounding.
Can ConvergePanel certify that a valuation is correct?
No. It can help compare assumptions across models and identify disagreement or internal inconsistency — it does not provide investment advice or certify a valuation. Building and defending a valuation model requires a qualified financial professional.
Explore related pages
ConvergePanel provides AI-assisted verification for informational purposes only. Not forensic analysis. Not legal evidence.
More in Research
Deep Research with Multiple AI Models
Run complex research questions through 5 AI models at once. ConvergePanel synthesizes consensus, disagreements, and bias signals into one structured brief.
Compare ChatGPT, Claude, Gemini, Grok, and Perplexity for Research
Compare ChatGPT, Claude, Gemini, Grok, and Perplexity for research. Learn when models agree, disagree, miss context, or need verification.
AI Research for Decision-Making Teams
Decision-making teams need shared, reliable research inputs. Multi-model AI surfaces consensus, disagreements, and uncertainty — not just one AI's take.