Hi Mitch To be blunt - I have little confidence in attempting to comment (feedback) or coach without any direct experience of what is on the table. The two single most important elements common to all NLP patterning: rapport and calibration of state (both agent of change and client) are fundamentally FA experiences and simply do not lend themselves to remote learning formats. I will offer a couple of possibilities and assume that you will act on them and offer comments on the differences encountered. You state that after 10-12 minutes, your client was in a flow state - here is one of the difficulties I referred to above. It would be at least tedious for you to attempt to describe verbally what the physiological indicators are and questionably effective. Similarly, your comments about watching the client very well to know when to gently push him into the context spatially anchored.  The structure (form and sequence) you report seem fine. Here are a couple of possibilities: in your instructions to the client, do you indicate in steps 1, 2 and 4 that he was NOT to consciously select anything different (desired state), simply notice what occurred (self-calibration and with respect to the stimuli that define the context selected). Did you ensure that the client found the auditory and visual stimuli in step 4 that define the context selected? - your description of the integration phase would indicate that you did.  You wrote, "However, two weeks later, I phoned him and he said nothing had changed and that he had gone to get a doctor who prescribed medicine for his 'disorder' - anxiety attacks just the day before I phoned - about 13 days after doing the alphabet game" Client's conscious reports ("he said nothing had changed"), while important to take into account, are best subjected to close scrutiny. I would suggest that you re-contact the client and schedule an appointment. When he arrives (check to determine whether he is on medication and its effects), use the verbal package, meta model, whatever to probe for differences. More interestingly and for the satisfaction of the objective that the client apparently has, develop with him a set of involuntary signals (minimally, ones that you can reliably read) and then secure responses to the questions that you have through those questions by calibrating the physiological responses to your questions. Once you have sorted out your own questions to your satisfaction, secure a signal from the client's unconscious that it will support the change and repeat the exercise taking into account what you learned from the responses to the questions above and ensuring that the client does not attempt to consciously select the differences (the desired state). To respond directly to your question - "Do the new code games always work if you do them properly?", first of all, be careful how you pose the question - the form you asked it in is tautological - the criteria for "properly" doing could be claimed to be that it works - hardly an enlightening criterion. Your case is the first reported to me that seem (by your description) to have followed (subject to my comments above) the sequence appropriately and failed to secure the results - congratulation?? At any rate, you are in possession of the only exception known to me - thus, the suggestions above for how to explore and learn from it while delivering value to clients. I am very interested as this is a new one for me and I am convinced that I learn more from the type of case you have than the typical success without the interesting wrinkles. John