High load average (ex. >0.7 or >0.8) indicates that the PostgreSQL server is nearing its limit for query workloads. Load averages over 1.0 indicate that queries are waiting to be executed by the server's CPUs and action should be taken to reduce server load or to upgrade the server instance.
Heroku defines PostgreSQL server load average as:
The average system load over a period of 1 minute, divided by the number of available CPUs. A load-avg of 1.0 indicates that, on average, processes were requesting CPU resources for 100% of the timespan. This number includes I/O wait.
A recommendation is created for your PostgreSQL server if the 5 minute load average is over 0.7 or the 15 minute load average is over 0.9.
Temporary spikes in load average may be caused by additional users using your service but may also be caused by poorly performing queries that should be improved.
Persistent, high load averages mean that action should be taken to prevent poor database performance from impacting your users. Tuning poorly performing queries (such as table scans) may help reduce load average, upgrading PostgreSQL's version may also help if you are running an older version, or upgrading the server instance size to one with more memory and CPUs will help reduce load average.