since your method is for requests payloads, you need to use memory-friendly  approaches. I don't know how often the method would be called, and how much requests at average your application would handle in daily bases. In all cases, it would be a good decision to assume the worst case scenarios, so handling them from the start would prevent from future high memory allocations. 

Also, I encourage you to use VS profiler or you could use [BenchmarkDotNet][1] in which would help give you a first-glance overview on your code diagnostic measurements. Use test samples to mimic the actual requests payloads, and also try to give at least three real test scenarios,  (one with least , one with average, and one with high API consumptions) just to diagnose and measure the overall performance (with memory allocation) of this addition.


The second approach is much better in logic and memory handling. The only missing part is a handling for `splitMaxLength` where it might be `<= 0` or `>= stringToSplit.Length`.

For readability part, `SplitStringByLastIndexOfAndMaxLength` I think if we use `SplitByLastIndexOf` it would be enough, since the method arguments would tell what type of input it's taking, and what it does returns. The only part is hidden, is the use of `LastIndexOf`.

I have modified your version to improve readability and code follow-up. and added some comments to explain some parts of the code.

    public static IEnumerable<string> SplitByLastIndexOf(string source, int blockSize, char delimiter)
    {
    	// in case of invalid source
    	if (string.IsNullOrWhiteSpace(source)) yield break;
    
    	// in case of invalid block size.
    	if (blockSize <= 0 || blockSize >= source.Length)
    	{
    		blockSize = source.Length;
    	}
    	
    	// reuse it rather than initializing it on each iteration
    	var removeSpace = 1;
       
    	// reuse it rather than initializing it on each iteration
    	var delimiterIndex = -1;
    
    	while (source.Length >= blockSize)
    	{
    		delimiterIndex = source.LastIndexOf(delimiter, blockSize);
    		
    		if(delimiterIndex == -1)
    		{
    			// if a word is larger than blockSize we still split it
    			delimiterIndex = blockSize;
    			// taking care to not remove a non space character
    			removeSpace = 0;
    		}
    
    		yield return source[..delimiterIndex];
    
    		source = source[(delimiterIndex + removeSpace)..];
    	}
    
    	// return any remaining text. 
    	if (source.Length > 0)
    	{
    		yield return source;
    	}
    }





  [1]: https://github.com/dotnet/BenchmarkDotNet