AlignCube sells a record you can check: every tool, the call on it, and the reason. So we keep the same record of ourselves.
Each entry below reached the live site or the product. It says what a person would have seen and what we changed, dated by the day it was fixed. New entries go at the top when they happen.
What you would have seen
During the seconds a deploy takes, a signed-in person could be sent to the sign-in page or left on a host error page, even though their session was fine. We found it on our own tab.
What we changed
A short server restart no longer counts as a sign-out. Reads are quietly asked again, and a reload notice appears only if the restart runs long.
What you would have seen
The paragraph under the plans was drawn over the last line of the paragraph before it, at every screen width, for about 12 hours.
What we changed
A negative top margin became a positive one, and a test now pins it.
What you would have seen
The homepage says a Cut needs evidence that a tool is unused, yet the published sample marked two tools Cut while their own reasoning listed checks still pending.
What we changed
The sample keeps one Cut that carries the evidence and moves the other to Review. A test runs the sample through the same evidence check the product uses.
What you would have seen
After we corrected the sample, the public board one-pager still told readers to cancel one tool before its next billing date, with no owner and no usage evidence behind it. It stayed that way for about 20 hours after the first fix.
What we changed
That tool is now Review on the one-pager, the marketing table and the decision-record strip, and a test checks that the one-pager and the app recommend the same action.
What you would have seen
The sample showed a yearly figure of $12,312 made by adding every line under review, open questions included, and presented it as realistic.
What we changed
The headline now counts only what the sample actually supports, and every sample surface shows the same figure.
What you would have seen
On shorter phone screens the cookie notice sat over the homepage's main button, so it could not be tapped. An earlier partial fix had not held.
What we changed
On short screens the notice now waits until after the first scroll.
What you would have seen
Once someone had an account, every Home link took them into the app, so the public site was out of reach.
What we changed
Those links now open the public site.
What you would have seen
The homepage sample said one tool overlapped another by about 80 percent, a precision the product cannot measure.
What we changed
The labels now name the tool that covers the job and state no percentage.
What you would have seen
A cost typed as $14,400/yr was read as $14,400 a month, twelve times too high, and flowed into the totals and the export. We have no record of a person hitting it.
What we changed
A cost that names its billing period is converted to a monthly figure, and a set of fixed test stacks now checks it on every change.
What you would have seen
For about seven weeks two of the four verdicts were the same amber, in the app and on the homepage, so they could only be told apart by reading the label.
What we changed
Each verdict now has its own colour everywhere a verdict appears, and a test fails if two ever share one.
What you would have seen
Our comparison page misdescribed one competitor's pricing, got another's service and features wrong, and misstated a third's position.
What we changed
Each line was checked against the vendor's own current pages, corrected and sourced.
Found something wrong that is not here? Write to [email protected] and it goes on this page once it is confirmed and fixed.