Note for buyers
Flakiness claims: how cases describe them and what to ask for
Research desk, testautomationvendors.com · published 2026-11-12 · data as of 2026-11-12 · 4 records
- 4
- flakiness claims
- 1
- with a before value
4 of 639 metrics mention flaky or unstable tests. 1 state a before value.
Counts come from verified records. A baseline is the before value a vendor states for a metric; quoted claims carry their grade and baseline flag.
- baseline:
- The measured value before the work started, against which a later value is compared. source
The claims
4 metrics use the words flaky, unstable or false failure.
- “Flaky tests cut from 18% to 4.6%”before 18% flaky test rate after 4.6% a1qa a1qa-1 grade B, partial baseline yes method yes
- “reduced test flakiness to under 2%.”before not stated after under 2% flakiness Codoid codoid-19 grade B, partial baseline no method no
- “Intelligent waits instead of hard-coded sleeps cut flaky failures by 85%.”before not stated after 85% fewer flaky failures ThinkSys thinksys-2 grade A, verified baseline baseline no method no
- “Flaky failures fell by more than 90%.”before not stated after more than 90% fewer flaky failures ThinkSys thinksys-3 grade B, partial baseline no method no
What to ask for
The flaky test rate as failures without a code change per 100 runs, before and after, over a stated number of runs.
Questions
How many cases mention flakiness?
4.
What is a flaky test?
A test that passes and fails on the same code without any change to that code.
Is flakiness scored?
Only as a metric under outcome verifiability, like any other metric.
What figure is most checkable?
A rate per runs, before and after, over a stated period.
Records used
This post reads 4 records; the newest verifiedOn date among them sets the data-as-of date.
Show record ids
More from the blog
Note for buyers
How to read a test automation case study: the fields that are usually missing
222 of 263cases omit start year
Field by field, how often 263 verified automation cases leave out team size, duration, baseline, client name and CI.
data as of 2026-11-12 · 263 records
Note for buyers
Why a percentage without a baseline is a claim, not a result
246percentage claims
246 percentage claims in verified automation cases, and how many state the starting value.
data as of 2026-11-12 · 126 records
Note for buyers
What an automation coverage figure can mean: the denominators seen in published cases
88coverage claims
88 coverage claims sorted by what they measure: test cases, regression effort, features or code.
data as of 2026-11-12 · 76 records
Related pages
Scores and grades describe published evidence as of the verifiedOn date on each item. They do not measure service quality. Verify with the vendor before contracting.