Tornado diagram for a decision score: which input matters most
The essentials in 30 seconds
- A tornado diagram is a bar chart that lists inputs vertically with the largest bar at the top (en.wikipedia.org, Tornado diagram).
- Each bar shows the score range when one input moves and the others stay at baseline.
- In the worked example, permits move the score by 0.131, nearly twice the swing of demand at 0.062.
- The baseline of 0.614 sits just under the EQOS Asia cut of 0.618.
- The EQOS Asia simulator lets you move each slider by hand.
A tornado diagram turns a sensitivity analysis into one picture that a committee reads in seconds. It ranks the inputs of a score by the swing each one causes. This guide shows how to build the diagram, how to read it against a threshold and where it stops being enough. It extends the guide on sensitivity analysis of a decision score, as of October 2026.
On this page
Compare the EQOS Asia offers and request a 30-minute pre-diagnostic.
What is a tornado diagram for a decision score?
A tornado diagram is a special type of bar chart in which the categories are listed vertically and ordered so that the largest bar appears at the top (en.wikipedia.org, Tornado diagram). The top bars are the items that contribute the most to the variability of the outcome. The shape, wide at the top and narrow at the bottom, gives the chart its name.
For a decision score, each bar belongs to one input or dimension. The bar runs from the score when that input is low to the score when it is high, with every other input held at its baseline. A reader sees at once which inputs deserve the most careful evidence and which ones can rest on rougher estimates.
How do you build a tornado diagram for a decision score?
Five steps build it. Compute the baseline score from the best available ratings. Choose a low and a high value for each input. Move one input to its low value, record the score, then move it to its high value, record again, and restore the baseline. Repeat for every input. Sort the inputs by the size of the swing and draw the bars, largest on top.
The tornado diagram is useful for deterministic sensitivity analysis, since the sensitive variable is modelled as having an uncertain value while all other variables stay at baseline values (en.wikipedia.org, Tornado diagram). A spreadsheet handles the repetition with one row per input. A figure lists the five steps.
What does a tornado diagram show in a five-dimension example?
Take five illustrative ratings: demand 0.8, talent 0.6, funding 0.7, permits 0.4 and partner 0.65, combined with an equal-weight geometric mean. The baseline score is 0.614. Move each rating by 0.2 in both directions. Permits swing the score from 0.535 to 0.666, a range of 0.131. Talent runs from 0.566 to 0.651, a range of 0.084.
Partner follows with a range of 0.077 (0.571 to 0.648), funding with 0.072 (0.574 to 0.646) and demand with 0.062 (0.580 to 0.642). The permits bar is nearly twice as long as the demand bar. The ranking tells the team where to invest effort: secure the permits first. The values are illustrations with equal weights, and they are not EQOS Asia results. The table lists the low and high scores.
How do you set fair ranges for a tornado diagram?
A bar reflects both the sensitivity of the score and the range chosen for the input. Using the same range of 0.2 for every input is a clean start, because it compares the effect of equal moves. A second version then uses ranges drawn from evidence: the lowest and highest plausible values for each input, with a source for each.
Show both versions when they rank the inputs differently. An input with a small effect per unit but a wide plausible range can still dominate the real uncertainty. Write the range next to each bar and give the reason in one line. A reader who sees the range can judge the bar, and the discussion moves to the evidence behind the range.
How do you read the bars of a tornado diagram against a threshold?
Draw the threshold as a vertical line. The EQOS Asia scale places the STOP and OK cut at 0.618 (EQOS Asia method page, as of October 2026). In the example, the baseline of 0.614 sits just below that cut, and the permits bar reaches 0.666, crossing it. The talent, partner and funding bars end between 0.646 and 0.651, short of the line.
Bars that cross the threshold mark the inputs that can change the verdict. Here, a better permit position could turn a STOP reading into an OK reading, while the others cannot do so within the chosen range. This is the practical use of the diagram: it separates the inputs that move the number from the inputs that move the decision.
What are the limits of a tornado diagram in a decision engine?
A tornado diagram moves one input at a time. In a real system, inputs rarely move alone. The EQOS Asia simulator shows an influence cascade, in which a dimension that moves drives the others through first-order and second-order effects, computed on a 9×9 matrix (EQOS Asia simulator page, as of October 2026). A one-at-a-time bar cannot show those chain effects.
Read the diagram as a first screen. It ranks the inputs by isolated effect, and a what-if analysis can then test combinations. The geometric aggregation adds a further point: because the score is a weighted geometric product, a collapsed dimension weighs heavily on the whole result (EQOS Asia case study pages, as of October 2026), and its bar will usually lead the chart.
How can you draw a tornado diagram with the EQOS Asia simulator?
Open the public Λ simulator and press Reset, which returns the 9 sliders to 50, a neutral state. Note the structural value, the figure the page calls the reference value to decide. Move one slider to 30, note the value, move it to 70, note again, and restore it to 50. Repeat for each of the 9 dimensions, then sort the swings (EQOS Asia simulator page, as of October 2026).
The Target button overlays the result of activating the top 3 levers, which gives a quick check of your diagram. For a sourced analysis, a 30-minute pre-diagnostic by video, scheduled in Singapore time (SGT), applies the equation to a real case at no commitment (EQOS Asia contact page, as of October 2026). The pricing page lists the tiers for a fuller study.
See also in this guide
Frequently asked questions
What is a tornado diagram?
It is a special bar chart where categories are listed vertically and the largest bar sits at the top (en.wikipedia.org, Tornado diagram). In decision work, each bar shows how far one input moves the score while the other inputs stay at baseline.
Why is it called a tornado diagram?
The bars are sorted from the longest at the top to the shortest at the bottom, so the chart narrows downward like a tornado funnel. The shape makes the most influential inputs visible at a glance and leaves the minor ones at the bottom.
Which input matters most in a tornado diagram?
The input with the longest bar. In the worked example, permits swing the score by 0.131, against 0.062 for demand. The ranking depends on the ranges chosen, so show both equal ranges and evidence-based ranges, and write each range next to its bar. A reader can then judge the bar and the evidence behind it.
Can I build a tornado diagram with the EQOS Asia simulator?
Yes, by hand. Reset the sliders to 50, move one slider to 30 and then 70, note the structural value each time, restore it, and repeat for each of the 9 dimensions (EQOS Asia simulator page, as of October 2026). Then sort the swings.
Test your knowledge
1. Where does the largest bar sit in a tornado diagram?
- At the bottom
- At the top
- In the middle
Show the answer
At the top.
2. Which dimension has the largest swing in the worked example?
- Permits
- Demand
- Talent
Show the answer
Permits.
3. What is the baseline score in the worked example?
- 0.618
- 0.614
- 0.666
Show the answer
0.614.
4. What do bars that cross the threshold mark?
- Inputs that can change the verdict
- Inputs with no effect
- Inputs without sources
Show the answer
Inputs that can change the verdict.
5. What value do the sliders return to after Reset in the simulator?
- 0
- 100
- 50
Show the answer
50.
About the author
Stéphane Jambu
President, EQOS Forecast Asia
Stéphane Jambu leads EQOS Forecast Asia and works as a consultant in e-commerce, search visibility, and AI. LinkedIn profile
Sources: EQOS Asia method page (eqos-asia.com/method.html), simulator page (eqos-asia.com/simulator.html), contact page (eqos-asia.com/contact.html), pricing page (eqos-asia.com/pricing.html), WeWork case study (eqos-asia.com/case-studies/wework.html), en.wikipedia.org/wiki/Tornado_diagram, consulted on 22 October 2026. Page updated on 22 October 2026.