Without using STRING_SPLIT and sticking only to SUBSTRING and CHARINDEX (lots of them), and assuming that your address data is normalized enough (e.g. all values for [address] will have four instances of ' ' and there are not going to be cases with Apartment Number or Streets with spaces in the name), the following code should split those addresses into three columns: Street, City, and County.
create table #temp (client nvarchar(30),address nvarchar(50))
insert into #temp values ('John Doe', '26 CHANCE ROAD SUTTON SURREY')
select SUBSTRING(address,1,CHARINDEX(' ',address,CHARINDEX(' ',address,CHARINDEX(' ',address)+1)+1)) as [Street]
--starts at beginning, length goes to 3rd instance of ' '
,SUBSTRING(address,CHARINDEX(' ',address,CHARINDEX(' ',address,CHARINDEX(' ',address)+1)+1)
,CHARINDEX(' ',address,CHARINDEX(' ',address,CHARINDEX(' ',address,CHARINDEX(' ',address)+1)+1)+1)-CHARINDEX(' ',address,CHARINDEX(' ',address,CHARINDEX(' ',address)+1)+1)) as [City]
--starts at 3rd instance of ' ', length is distance from 3rd instance of ' ' to 4th instance
,SUBSTRING(address,CHARINDEX(' ',address,CHARINDEX(' ',address,CHARINDEX(' ',address,CHARINDEX(' ',address)+1)+1)+1),20) as [County]
--starts at 4th instance of ' ', length just needs to go through to the end
from #temp