Scan stability - Retry on failure or On fail, record last known good?

[cybershambles]

"name": "Crikey", "url": "www.crikey.com.au", "wayback": ["X", "X", "X", "X", "X", "X", "X", "X", "X", "X", "X", "X", "X", "X", "X", "X", "X", "X", "X", "X", "X", "X", "X", "X", "X", "X", "X", "X", "X", "X", "X", "X", "X", "X", "X", "X", "X", "X", "X", "X", "B", "X", "B", "X", "X", "X", "X", "X", "X", "B", "X", "X", "B", "B", "B", "X", "B", "X", "B", "B"],

Because we're running a lot of scans - I've noticed lots of X's in the daily scans.

1. Retry the failures until the scan is successful
   or
2. If the last known score was B, but it's failed today... keep the last known as B until the next successful test which might lower the score.

Having all of these X's will cause false negatives, stopping the change updates from being notified in our twitter.

The graph on our site is looking messy because the graph keeps dropping and returning.