Sensitivity Testing explained: this guide covers what it means, who it applies to, the step-by-step process, documents required, fees, due dates and penalties in India — so you can stay compliant with confidence and avoid costly mistakes.
A sensitivity table is only as useful as the range it tests. Too narrow and it understates uncertainty; too wide and the results have no practical relevance.
The four good-practice rules
- Perform sensitivity testing for the two or three assumptions that have the greatest impact on value.
- Test ranges should represent realistic alternative outcomes rather than extreme or arbitrary assumptions.
- Results should be presented in terms of the effect on value, not merely the change in the underlying assumption.
- Assumptions exhibiting high sensitivity should receive enhanced justification and disclosure.
Illustrative test ranges
| Approach | Assumption | Illustrative range |
|---|---|---|
| Income (DCF) | Discount rate (WACC) | 100 to 150 basis points |
| Revenue growth rate | 2 percentage points | |
| EBITDA margin | 100 to 200 basis points | |
| Terminal growth rate | 50 basis points | |
| Cost (Asset) / liquidation | Asset-specific haircuts | 10% |
| Sale horizon / realisation timeline | 6 months |
These are illustrative and should be adapted where the circumstances indicate a different range would be more appropriate. The objective is not to test extreme outcomes but to show the effect of reasonably possible changes.
The terminal growth range is 50 basis points — the tightest of the four DCF ranges, and a third of the WACC range. That is not because terminal growth matters less; it matters disproportionately. It is because a realistic alternative for a long-term growth rate anchored to economic growth is genuinely narrow. A half-point either side of a sustainable rate is a real possibility; two points is not a sensitivity, it is a different assumption.
The liquidation ranges work the other way. A 10% variation on an asset haircut is wide in absolute terms, and appropriately so — the evidence base for a forced sale discount is thinner than the evidence base for a government bond yield.
A range chosen for symmetry across the table rather than from what is realistic for that input is arbitrary, which is one of the recognised failure modes.
What the analysis should cover as a minimum
- DCF valuations — ordinarily, at a minimum, discount rates, revenue growth, EBITDA margins and terminal growth rates.
- Liquidation value analyses — ordinarily asset realisation assumptions, key asset-specific haircuts and sale horizon assumptions, as these are typically the principal drivers of value.
The two-way matrix
An enterprise value matrix varying WACC down one axis and terminal growth across the other — for example WACC from 11.0% to 13.0% against terminal growth from 5.0% to 7.0%, with the base case at the centre — makes the compounding visible in a way that two separate one-way tables cannot.
Such matrices allow users to assess the extent to which value depends on key assumptions and to identify assumptions that warrant particular scrutiny. In the illustrative matrix, the corners differ by roughly a factor of two across ranges that are each individually modest — which is precisely the point being communicated.
Presenting the effect, not the input
The third rule is the one most often broken. A table stating that WACC was tested at 11.0%, 12.0% and 13.0% tells the reader nothing they could not have guessed. A table stating the enterprise value at each of those rates tells them how much the conclusion depends on getting the discount rate right — and prevents stakeholders from assessing materiality if omitted.
Sensitivity testing feeds back into disclosure
The fourth rule closes the loop: where an assumption turns out to be highly sensitive, it moves into the category requiring enhanced justification and disclosure. Sensitivity testing is therefore not only a communication tool but a triage mechanism that tells the valuer where the report's justification effort belongs.
Common mistakes
- Applying the same percentage band to every assumption regardless of what is realistic.
- Testing five or six inputs instead of the two or three that drive value.
- Tabulating assumption changes without the corresponding value.
- Finding an assumption highly sensitive and leaving its justification unchanged.