Other indicators can include malware sample hashes or actor-specific detection rules (Example: YARA,Snort or Netwitness rules).
These indicators are typically not public. Some can be accessed if you pay the right sum of money and undergo vetting,still,some are kept private within the relevat security firms or organizations.
As you can imagine,Google has their hands in many pies including security research and threat intelligence collection(Everyone loves their VirusTotal intelligence product). They can scan email metadata for any of these indicators as they see fit.
Generally speaking,some indicators are of such high quality, they can be used to detect well crafted spear phishing by a nation state actor. But most are good only to detect untargeted attacks or targeted attacks that include a large number of targets.
Hope that answered your question.
Geo-political intelligence analysts and sometimes field officers (aka spies) also play a role. For example,if a specific journalist was attacked while working on an article critical of a specific country and at least one indicator was traced to an a real human associated with that country's spy agency the a confident attribution can be made.
The term "nation state" does sound a bit fancy but it is different from "country" in that many nation states do not act on behalf of their people or to protect the interests of their conuntry. If you use "country" that would include the people of the country while "nation state" simply means the organization running the country. Plus you have situations like taiwan,hong kong and chechnya where those nations are effectively their own nation state but they are under the control of a different country.
Here is a good reference attribution of APT3 (Chinese Ministry of State Security (MSS)): https://threatpost.com/apt3-linked-to-chinese-ministry-of-st...