Take the profile Maheu-10 as an example.
As of this moment as I write this post I look at the profile and see there are 3 unmerged matches: Maheu-31, Maheu-26, and Maheux-6 - these 3 profiles are all already related to one another.
Now Maheu-10 has pretty scant details to work with, but I take a stab at it anyway. I pull open a relevant source, such as:
Tanguay, Cyprien. Dictionnaire généalogique des familles canadiennes depuis la fondation de la colonie jusqu'à nos jours. vol. 5 sect. 2 : Lei-Mer; page 455, Québec, Canada: Eusèbe Senécal, 1871-1890. Library and Archives Canada, Ottawa, Canada. [Digital Image] Ancestry.com
In this source I find:
1. Pierre (son of Jean Maheu and Michelle Chovin) b.1634, m.1659 Jeanne Drouin (Maheux-6)
2. Charles (son of Pierre Maheux and Jeanne Drouin) b,1666, m.1688 Marie Charlotte Garnier (Maheux-5)
3. Pierre (son of Pierre Maheux and Jeanne Drouin) b. 1669, m.1691 Marie-Louise Garnier. (Maheu-31)
4. Pierre (son of Charles Maheu and Marie-Charlotte Garnier) b.1693, m.1715 Louise Giroux (Maheux-4)
5. Pierre (son of Pierre Maheu and Marie-Louise Garnier) b.1694, m.1715, Suzanne Giroux (Maheu-26)
6. Pierre (son of Pierre Maheu and Louise Girou) b 1717, m.1739, Antoinette Guevremont. (Maheu-10)
I can now say that all the unmerged matches on Maheu-10 should be rejected, and, as it happens, I can also say that Maheu-10 should be connected as the son of Maheux-4. Had the additional research instead indicated that Maheu-10 was in fact the same person as one of the other profiles I would have proposed merges instead of rejecting the matches.