Visual Tools
Calculators
Tables
Mathematical Keyboard
Converters
Other Tools

How to Find a Geometric Probability

A geometric variable is the number of the trial on which the first success happens. The first success on trial kk means k−1k - 1 failures in a row and then a success, so each probability is a single product; "more than kk trials" means the first kk all failed.

Recognize the task

Independent trials with the same success probability are repeated until something happens for the first time, and the question is about when: "roll a die until the first six", "keep calling until someone answers", "P(first defective is the 5th part)P(\text{first defective is the } 5\text{th part})". Test: there is no fixed number of trials and you stop at the first success. Then X∼Geom⁡(p)X \sim \operatorname{Geom}(p) with values k=1,2,3,…k = 1, 2, 3, \ldots
The model is defined in the geometric distribution lesson; this page counts trials, not failures.

Two ways to do it

PMF for an exact trialexact
P(X=k)=(1−p)k−1 pP(X = k) = (1-p)^{k-1}\, p
Use the geometric PMF: k−1k - 1 failures, then the success.
Tail formula for rangesat most / more than
P(X>k)=(1−p)k,P(X≤k)=1−(1−p)kP(X > k) = (1-p)^k, \qquad P(X \le k) = 1 - (1-p)^k
The first kk trials all fail; the geometric CDF is the complement. No sum needed.

Steps

StepOn a fair die rolled until the first six, P(X=3)P(X = 3)
1Check the setup and name pp. Independent trials, the same success probability, stop at the first success. XX is the trial number of that success.p=16p = \frac{1}{6}, 1−p=561 - p = \frac{5}{6}, X∼Geom⁡(16)X \sim \operatorname{Geom}(\frac{1}{6})
2Write the event in terms of XX. "On the kkth roll" is X=kX = k. "Within kk rolls" is X≤kX \le k. "Needs more than kk rolls" is X>kX > k."first six on the third roll": X=3X = 3
3Apply the PMF or the tail formula. An exact trial uses (1−p)k−1p(1-p)^{k-1}p. A range uses (1−p)k(1-p)^k for the tail, or 1−(1−p)k1 - (1-p)^k for "within kk".(56)2⋅16=25216\left(\frac{5}{6}\right)^2 \cdot \frac{1}{6} = \frac{25}{216}
4Simplify and round. Keep fractions exact if you can, then round at the end.P(X=3)=25216≈0.1157P(X = 3) = \frac{25}{216} \approx 0.1157

Worked examples

Example 1 · an exact trial

PMF
p=0.2p = 0.2 per call, P(first sale on call 4)P(\text{first sale on call } 4)
X∼Geom⁡(0.2)X \sim \operatorname{Geom}(0.2)
P(X=4)=(0.8)3⋅0.2P(X = 4) = (0.8)^3 \cdot 0.2
three failures, then a success
P(X=4)=0.1024P(X = 4) = 0.1024

Example 2 · success within k trials

CDF
p=0.2p = 0.2, P(a sale within 4 calls)P(\text{a sale within } 4 \text{ calls})
the event is X≤4X \le 4
P(X>4)=(0.8)4=0.4096P(X > 4) = (0.8)^4 = 0.4096
the first four calls all fail
P(X≤4)=1−0.4096=0.5904P(X \le 4) = 1 - 0.4096 = 0.5904
complement

Example 3 · after a run of failures

memoryless
p=0.2p = 0.2, three calls failed; P(X>5∣X>3)P(X > 5 \mid X > 3)
P(X>5)P(X>3)=(0.8)5(0.8)3\frac{P(X > 5)}{P(X > 3)} = \frac{(0.8)^5}{(0.8)^3}
X>5X > 5 already implies X>3X > 3
=(0.8)2=0.64=P(X>2)= (0.8)^2 = 0.64 = P(X > 2)
past failures do not count
Check your numbers in the discrete distributions calculator.

See it happen

Every trial that fails multiplies the remaining probability by 1−p1 - p, so the bars of the PMF shrink by the same factor from one trial to the next: a geometric sequence, which gives the distribution its name. The tail (1−p)k(1-p)^k is the probability that the first kk bars were all skipped. See the decay in the geometric distribution visualizer.

Common mistakes

One failure too many

✗ P(X=3)=(56)3⋅16P(X = 3) = \left(\frac{5}{6}\right)^3 \cdot \frac{1}{6}
✓ trial 33 is the success, so only 22 failures: (56)2⋅16\left(\frac{5}{6}\right)^2 \cdot \frac{1}{6}

Mixing the two conventions

✗ counting failures before the success, then using (1−p)k−1p(1-p)^{k-1}p
✓ with YY = failures, P(Y=k)=(1−p)kpP(Y = k) = (1-p)^k p and Y=X−1Y = X - 1; state which one you use

Wrong exponent in the tail

✗ P(X>4)=(0.8)3P(X > 4) = (0.8)^3
✓ more than 44 trials means the first 44 failed: (0.8)4=0.4096(0.8)^4 = 0.4096

Thinking a success is "due"

✗ after 33 failures, P(X=4∣X>3)P(X = 4 \mid X > 3) is larger than pp
✓ it equals p=0.2p = 0.2; the trials are independent (memoryless property)