I've likewise on occasion been notified of DNA matches where the
cumulative cM value wasn't particularly high, but with one very long
matching segment... plus no apparent mutual ancestor identifiable in
their/my records-based tree.
In these instances, my "guess" was that this could be attributed to
endogamy; that is cousins or uncle-niece marriages in the family
history resulting in a composite segment issuing >from intermarriage
between more than one intermediary ancestor, but originally having
its source in the still earlier mutual ancestor.
In other words, the original ancestral segment was re-created >from
partial components of the mutual ancestor's DNA which transited down
the generations separately, and then were recombined again.