I can't replicate 'auto complete' locally

  • 14 October 2023
  • 4 replies



I’m trying to replicate the ‘auto improve’-functionality of Testim with my locally based web-app (runs only on my computer without docker). I simply can’t replicate it.  I ran my tests starting from localhost as the URL via the local CLI. I then changed locators and the overall score dropped to 55% of the Target Element. Wouldn’t that mean that it should auto- improve? But it doesn’t have ‘AI’ standing anyway..?  Also what does the -1 mean on the screenshot bellow?

Thanks! I 


Best answer by Meital 16 October 2023, 10:46

View original

4 replies

Userlevel 7
Badge +2

@alex_read@Meital, @Matan is it something you could help with? 


Hello, Daniiii, and thanks for reaching out! Auto Improve is available at the Pro tier for Testim, so if you're interested in upgrading, please reach out to to extend your trial and get in touch with one of our Solutions Consultants. Your trial also changed into a Community Plan when it ended, which allows you to continue using Testim with limited functionality for free. You'll find self-service help documentation like this one about dedicated run tunnels that may help you tackle your project from a different angle.

Userlevel 5
Badge +3

I’ve not been an actual user of TestIM myself, so could only provide some general tips, but maybe, hopefully, one of these might be useful :c) To your first point, from a quick scan of the TestIM site, I found a note that it should auto-improve if the score is less than 70% by default.


  • It seems the auto-update needs to be switched on to function:
  • Perhaps the 55% is the best that the tool can get to - i.e. if it’s using the DOM to find bodyDiv/form/input-of-type-text or find the textbox directly underneath & relative to the one named “userNameTextbox”. Are the elements you want located, coded with an ID or name property? as this would likely bump up the locator to 100%

Some other more general notes:

  • Other similar tools have a threshold setting, whereby you can alter the 70% and say, for example, only auto-improve elements with a score beneath 10%. Perhaps TestIM has such a setting & it might’ve been altered?
  • There is a red error warning just offscreen in your screenshot. What does this say please?
  • If you hit the edit link of that 1st Id (nameSubmitBtn), what shows under there please?

The -1 means the TestIM team may have done a poor job of developing this tooltip and testing it. It’s not visually easily meaningful to end-users and I find no mention of such a value when performing a search online. I would say just ignore it for now. Perhaps later if you have a full manual locally describing the tool usage, or need to contact TestIM’s own support team, it might tell their team some information for internal troubleshooting...


thanks for your detailled reply, alex_read! As for your first point: I did switch on the ‘Show improved steps’ as you can see on the screenshot. As far as I’ve understood it from my online research, the 55% means that Testim is 55% certain that this is the target element. The elements have coded ids. Testim recognizes that the id has changed (there was also a tooltip stating that.) 

I do have several days of a free trial left. And for many days I’ve ran the test suite over the local CLI with my local chromeDriver, but I still can’t see an auto improvement. And yes, the meaning behind the -1 I also can’t find online. 

Thanks anyways! I just don’t understand why I can’t reproduced it. I question myself, if the auto improvement only works on tests that don’t detect bugs in the software… Or the reproduction of the functionality only works on scheduled tests…? Or it just doesn’t work on localhost-based tests? Does anybody else know perhalps?