Any localized place information you have is probably the best clue you will ever get.
Someone contacted me recently for help. An ancestor of his had been attached to the wrong family on the FamilySearch tree. He needed to identify the actual parents and provide proof so others would stop perpetuating the mistake. I looked at the family group and instantly understood what he was getting at: the attached parent couple was well documented living over 200 km/ 150 miles as the crow flies (and maybe twice that by the land routes of the time), from the somewhat isolated Ontario township where his own ancestor was known to be living. While both families were definitely in Ontario, another researcher, probably without much understanding of Ontario geography, assumed incorrectly that meant they were in close proximity.
Ontario is easily twice the size of most major European countries. You could basically fit both Texas and California into Ontario. While Ontario historically did not have the population density of a European country, actual physical distances still matter. In the early days, distance mattered quite a bit.
Knowing the surname and which township the actual parents must have been living in provided the key starting point to solving this problem. A digital index revealed one person with that surname who had letters left at the post office servicing the township in the 1820s. Further searches revealed there were absolutely no vital or church records existing for this person. However, given the township, that is not really surprising if he died before 1861 (generally the 1851 censuses survive in Ontario, but not for this place). The church and burial record survival for this area is known to be poor.
Further searching identified six other people with the right surname who researchers associated with this township, potentially siblings of the target brick wall. However, two of those people were revealed to be incorrectly assigned to the township and, once their authentic locations were identified using vital and census records, a bit of further research identified their parents and actual family units. That still left four potential siblings. A noticeable commonality between these five people was their longevity. This proved to be a disadvantage because when they died in their late 80s and 90s, the grandchildren and in-laws who reported the deaths did not know the names of their parents. It was becoming clear why these people had endured as brick walls in so many trees.
While initial searches revealed there was no handy family surname file of clippings and 19th or 20th-century research from incidental sources, such files did exist for two families the potential siblings had married into. These files, in the ever-useful Herbert Clarence Burleigh fonds at Queens University (scanned on Internet Archive), contained crucial sketched family trees and one mysterious typed manuscript genealogy.
Both files indicated all five people under investigation were siblings, and they were indeed children of the tardy letter collector. And, crucially, their mother proved to be the daughter of a well-documented Loyalist. The extensive typed genealogy included tantalizing references to information from 17th or 18th-century Dutch bibles and phrasing indicated information collected from living descendants, although it was not clear when these activities occurred and specifically who initiated it.
Following up on a reference to one bible-owner, who had a very rare surname, further untangled the situation and revealed that the typescript was a partial copy of an article series that appeared in the New York Genealogical and Biographical Record in 1882 and 1890. The original article (accessible via Google Books and FamilySearch) revealed the NY-based author had solicited descendancy information for a notable NY Dutch family. Given the intermarriages and the author's willingness to include female descendant lines when known, this had captured two elusive brick-walled families in the isolated Ontario township and a related brick-walled family in Ohio.
The information was only recorded because the 1880s researcher's mail campaign reached two unmarried siblings born in the 1820s who still lived on the original Ontario family farm. They responded with everything they knew and then both died in 1891.
With such treasured information in hand, I corrected the initial mistaken parentage and set about fixing other indicated mistakes on the FamilySearch tree. I always hesitate to disconnect an assigned child unless I have reasonable evidence the link is wrong and when doing so, I try to reassign the child to correct parents in order to avoid disappointing any descendants.
When I hit one such case, a person found on the tree, but not in the published genealogy, I instantly understood why the (probably erroneous) connection had been made. One of the children had been given as their first name, the somewhat unusual Dutch surname of their currently assigned grandmother. I can imagine the researcher's enthusiasm in making this connection (surely this is good circumstantial evidence!). However, in reviewing and further assembling the census records for this family it became apparent that the birthplace for the person incorrectly attached to the Ontario family was not in 1820 "Canada" as listed in the profile (although his wife was Canadian), it was, per the 1875 NY state census, Herkimer County NY.
Within five minutes of investigating the limited number of appropriately surnamed families in Herkimer in the 1820s, I found another couple, married in Herkimer in 1807 ... and the wife had the same somewhat unusual Dutch surname ultimately bequeathed to her grandson. By coincidence, two such couples existed at roughly the same time, one in Ontario and one in New York.
The researcher who made the incorrect association would have found this other couple before ever looking at the Canadian family, if they had just been looking in the right place.