Title: Reliability Issue with Long-Term Data Tracking in ChatGPT – Concern Before Paid Subscription
I have been using ChatGPT for several months as a structured log for my health and fitness progress. I consistently shared daily information such as walking distance, step counts, weight readings, sleep data, and blood sugar reports. Each time I asked whether the information was being recorded, the system confirmed that it was.
However, when I later requested analysis or summaries of my data, a significant portion of the information I had previously shared was missing from the reconstructed records. In some cases, only a fraction of the logged information appeared in the analysis.
From a user perspective, this creates a serious reliability concern. When a system confirms that information is being recorded, the expectation is that it will remain available for future analysis.
The main problems I experienced were:
• Information logged across conversations is not reliably accessible later.
• Daily logs can disappear from summaries even when they were shared.
• Uploaded files and screenshots may expire, making earlier records inaccessible.
• The system can generate analysis based on incomplete data without clearly indicating that the dataset is partial.
For users attempting to track long-term information such as health metrics, this creates frustration and undermines confidence in the analysis.
Another concern relates to the subscription model. Users are often encouraged to use the system over time before committing to a paid monthly subscription. If the first months of usage already create uncertainty about whether the system reliably records and retrieves previously shared information, it raises an important question:
Why encourage long-term usage leading into a subscription if the system cannot consistently maintain the integrity of the information shared during that period?
Suggestions for improvement:
-
Provide a true persistent logging feature for structured data such as fitness or health tracking.
-
Ensure continuity of memory across conversation threads for the same user.
-
Prevent expiration of files that are part of ongoing records.
-
Clearly indicate when analysis is based on partial data rather than presenting it as a complete record.
ChatGPT is very useful for discussion and interpretation, but in its current form it is not reliable as a long-term structured log unless users manually consolidate their data elsewhere.
Improving data continuity and reliability would significantly increase user trust and make the platform more suitable for long-term use.