Perform Action @ position relative to object

You can certainly do what you're describing using a script.  For instance, you can get the top left screen coordinate of the label and the width of the label.  Adding those values together will give you the right side of the label in screen coordinates.  Add 50 to that and you should have the screen coordinate value you want.  Then you can click there (e.g. Sys.Desktop.Click <Add your coords here>).

However, I can't endorse using that approach.  You should be able to find that edit box reliably by creating a name mapping using the right object mapping properties.

What properties do you see for it in the Object Spy?  There has to be something there that will work.  Caption?  ObjectIdentifier?  Name?  ID?

Imagine the Object Browser shows the following hierarchy structure, based on the form

If the the text and the textbox is within the same container, then it's possible to get the parents' child. For example, I want to get the label object, which doesn't have a unique id. Therefore, I'm going to use the textbox (since it has a unique id), to get the parent. I'll use the parent to get the child object, label.

 var firstname = obj.Frame("iframeResult").Form(0).Panel(0).Textbox("fname");

 // Textbox("fname") parent is obj.Frame("iframeResult").Form(0).Panel(0)
 parent = obj.Frame("iframeResult").Form(0).Panel(0).Textbox("fname").Parent();    

 var label = parent.Label(0); // Same as obj.Frame("iframeResult").Form(0).Panel(0).Label(0)

You might be able to do something similar.

In this example, I have grouped the text and textbox into a grid and have provided an AutomationId called "LoanAmount". You can easily get the required object, if your developers implement the controls for automation.

"Why? Because the order of these rows moves around a lot (as developers design) and I always want to be able to just find the textbox by locating it relative to it's corresponding text located to it's left."

If dev does not name the objects effectively to make the application 'automation friendly / testable' it makes this task much harder. 

If you can find and map or script an instance of the label object  or textbox object, the task will be pretty easy.  At this level you are independent of the screen position and just locating the object.   

If you still want to perform actions relative to the position of the label, you just need to do a little math based on the position of the found label object. 

The object properties should hold the position of the found object...

Also as described by others you can use the parent or child property of the found object.  If both exist in a single stack panel you should be able to move up a level or two and then search down from that parent to a target child which should be unique within that part of the app tree.  Again with this approach the onscreen position of the object is irrelevant as you are working with an instance of the object using it's properties. 

Otherwise... would OCR work for a desktop app in this situation?

https://support.smartbear.com/testcomplete/docs/testing-with/object-identification/ocr/index.html#about

Snip from the doc article

If there are several text fragments that match the text, specify how the operation will choose the needed text fragment:

• Nearest to center
• Left most
• Right most
• Top most
• Bottom most
• Largest
• Smallest
• None - The operation will simulate a user action on the first fragment it finds when searching from top to bottom and from left to right, and will then post a warning to the test log.

3. Specify Method

On the Specify Method page, select the user action you want to simulate:

• You can simulate common actions TestComplete provides for all on-screen objects. You can find their description in the TestComplete Help.

• To simulate a user action next to the selected text fragment, select one of the following methods:

  • ClickNextTo (for desktop and web applications)

  • TouchNextTo (for mobile applications)

  • SendKeys

I thought 'On-Screen Action' would be the way to go. But like said, seems like this only works on the selected object. Not on an object relative to the selected object. 

Thank you everyone for all the help. After reading what everyone had to say, I could see there was no way to do this via keyword features so I scripted it out. Thanks again for all the help figuring this out. It would be nice to be able to just find an object relative to another object via a keyword feature; oh well. I tried the OCR Actions and it too seems to only work within the object that you extracted the text from (yes, poor naming or no naming makes things A Lot harder).

function testing(rowLabel)
{
  // --- 1. Declaration section
  //var rowLabel = "Name"  
  var table = Aliases.Viewport.ViewportWindow1.ViewportWindow2.SchematicWindow.zshowDataGrid;
  
  // --- 2. Search the table for the row contianing the text, 'rowLabel' in the Name column
  var propertyNames = new Array("ClrClassName", "Text");
  var propertyValues = new Array("TextBlock", rowLabel);  
  var labelTextBlock = table.FindAllChildren(propertyNames, propertyValues, 6);  
  
  // --- 3. Get the parent row for this item
  var rowContainingLabel = labelTextBlock[0].Parent.Parent.Parent.Parent;
  
  // --- 4. Get Cell in column 2 found in this row
  var propertyNames = new Array("ClrClassName", "WPFControlOrdinalNo");
  var propertyValues = new Array("DataGridCell", "2");    
  var cellInColumnTwo = rowContainingLabel.FindAllChildren(propertyNames, propertyValues, 3);
  
  // --- 5. Get the textblock in this cell  
  var propertyNames = new Array("ClrClassName", "WPFControlName");
  var propertyValues = new Array("TextBlock", "_textBlockValue");    
  var textBlockCorrespondingToLabel = cellInColumnTwo[0].FindAllChildren(propertyNames, propertyValues, 7);
  
  // --- 6. Finally get the text (or set the text if needed) for the field
  var value = textBlockCorrespondingToLabel[0].WPFControlText;
  
  // --- 7. Return retrieved value if needed
  return value;
}