Gong records sales calls, transcribes them, and runs analysis over the result to flag deal risk, surface what buyers are actually saying, and give managers visibility into conversations they were not in. Most organisations that buy it use maybe a quarter of it, typically as a recording archive plus a dashboard somebody checks weekly, which is an expensive way to buy a search index. The Academy exists to fix that and it is reasonably good at it. The role-based structure is the right call and properly executed.
What an account executive needs from Gong is not what a frontline manager needs, and neither resembles what an admin needs. The AE path is about preparing for calls, reviewing your own conversations and acting on deal signals. The manager path is about coaching from evidence rather than anecdote, which is the actual promise of the product, and about not turning call review into surveillance theatre. The admin path is configuration, integrations and permissions.
Splitting these rather than shipping one generic product tour respects people's time and increases the chance any of it sticks. The AI Trainer material is the part I found most interesting, because it addresses a real gap. Sales skills improve through practice under pressure and almost nobody practises, because the only available practice is live calls with actual prospects, where the cost of fumbling is a lost deal. Roleplay with a manager works and does not scale and is mortifying enough that people avoid it.
An AI roleplay partner that will play a sceptical CFO for the fifteenth time without getting bored solves a genuine problem, and the training around it covers how to set up scenarios and use the feedback. Whether the simulation is good enough to transfer to real conversations is an open question I cannot answer from the course material, and the instinct is correct. Two substantive criticisms. First, the training teaches you to trust Gong's signals without teaching you when not to.
The platform will tell you a deal is at risk, that a competitor was mentioned, that sentiment shifted. Those outputs come from models operating on transcripts, and they are wrong sometimes, in patterned ways. Sentiment analysis on a transcript misses tone entirely. Risk scoring learned from historical deals will be confidently wrong about a deal shaped differently from the training distribution.
A manager who treats the risk flag as ground truth will coach the wrong rep about the wrong thing, and any honest course about a prediction tool would spend twenty minutes on its error modes. This one does not. Second, and more uncomfortable, there is very little here about the ethics and legality of what the product does. Gong records conversations, and recording consent law varies by jurisdiction in ways that genuinely matter, particularly across European markets and between US states.
Beyond the legal minimum there is a culture question about what it means that every word a rep says to a customer is retained, analysed and visible to their manager's manager. Teams handle this well or badly and the training could shape that, and instead it treats recording as an unremarkable default. That is a meaningful omission in training aimed at the people setting these norms. The access limitation is straightforward.
This is customer training. Some of the catalogue is browsable, the substance assumes you are signed in with a workspace containing real data, and none of it is much use without the product. Three point seven. Well-organised, genuinely helpful for underusing customers, credulous about its own outputs, and silent on the questions its product raises.