1. The Hawthorne Effect - any change is likely to improve response, just because you changed. It may not be a response to the particular change itself.

2. The long-term training of readers. Training them to decide whether to click through or not based on what you write in the first bit may make it difficult in future.

3. All the points you identify for the full post, still apply to some extent for part posts.

How about a standard vanilla announcement, but with occasional variations to keep people on their toes.

Ahh, the limits of scientific testing