The results reported by Zhang show a high quality of experiment design and data collection. However, their measurements show just 1.5 watts excess thermal power at 150 watts heater input and 2.5 watts at 250 input, with a maximum of 5 watts seen. That is a COP of 1.01-1.02, which I would not call encouraging.
The term "COP" is meaningless in cold fusion. Actually, it only applies to heat pumps, which have no relevance. The input power is not transformed in any sense. It does not trigger output. There is not fixed ratio between input and output. Input power in this experiment serves only to raise the temperature. It could easily be reduced by improving insulation, but that would be inconvenient and possibly dangerous. There is no difficulty measuring input power with extremely high precision, so whether it is 10 W or 300 W, it does not reduce the s/n ratio, and it makes no difference at all.
It is a shame absolute output power is not higher in these experiments, but input power has nothing to do with it. What is more important from a scientific point of view is that output can be measured with confidence, and it can be generated reproducibly. That is extremely encouraging. Without progress in reproducibility, no other experimental, scientific or ultimate practical progress can be made.
High output is needed for commercial products, but cold fusion is light years away from commercial or practical use. So, the criteria of commercial reactors do not apply, any more than they did to the Curie's first radioactive sample, or the Chicago Pile 1, or to Tokamak reactors today. This device has only scientific value. It can only be made into a practical device by spending hundreds of millions of dollars more in R&D at dozens of labs.