In every adventure, the GM's greatest enemy is an unexpected event that sends things astray. During an investigation, dice can easily create an odd result interpretable as a false lead precisely when that is disruptive. With an open roll, the player may see the dice behind the unusual event, but the character has found no explanation. How should we assess this? It is easy to say unusual results and almost inexplicable events belong to the genre and should be welcomed. But there are at least as many arguments for saying that a bad roll puts something unusual—a potential clue—where we do not want it and can derail the adventure.
Consider a skilled tradesperson's chance of succeeding at an easy task. With a reject rate like that, they could not work anywhere. That is why you should not call for a roll when there is no exceptional pressure and the character is doing something within their competence. The probabilities would not be realistic. Wherever you call for or make a roll, the situation and odds are already outside the ordinary. These are unusual events whose unusual nature the characters will ideally investigate.
Worse, they may grow accustomed to your world being full of unexplained, unusual elements and fail to notice when a piece of evidence or important clue stands out. An interesting possibility is preparing for strange situations produced by bad rolls. Decide where investigation will lead if the group looks into that or anything else. You say the players will not investigate? Explain the bad roll in a way that makes them investigate the explanation, turning a risky situation to your advantage.
If a roll does not produce the strange scene and no other oddity leads in the desired direction, this remains available. If characters go astray by misinterpreting evidence or a clue without rolling, use the fact that Call of Cthulhu contains things the average person would call supernatural. It might be a vision or an unusually strong intuition. A failed roll remains a failure in its particular situation, but can lead to different evidence in the same way. Try explaining strange results so they build atmosphere and lead to evidence.
That evidence is not free: the supernatural experience may reduce Sanity, it is hard to cite the knowledge gained, and finding a suitable natural explanation for events is difficult. The apparently free help received upon failure is therefore not necessarily advantageous. There is no free lunch, and in this respect the help a frequently rolling player receives on failure is expensive.