Hi,
to interpret the coherence value, one should give many more details: what model you are using, what parameters are estimated, what options you applied, which images graph, whether weights are used, whether the atmosphere is corrected or not, and so on and so fort, the list could be long. Without having those details, you cannot interpret well coherence values.
However, in your case, whatever the options you choose, since you have only 3 images, and since you need a reference to analyze the phase, it’s like having only 2 samples. whatever model you try to fit, it will always match your data, residual will be zero, and the coherence will be one.
best