What this site returns stands on one assumption: that both types are correct. Move the assumption and the conclusion moves with it. So before anything else, here is how hard typing is, and how to use the result.
Each of the 16 relations is decided by laying two types over each other. So the certainty of the result is the certainty of your own typing multiplied by the certainty of theirs. If either one is wrong, a different relation comes out.
One letter can flip the result to its opposite. Seen from ENTP, an ISFJ is conflict and an ISFP is duality. Two neighbouring types, opposite ends of the sixteen. So when a result feels wrong, doubting the typing is usually faster than doubting the description.
Answering questions about yourself, you tend to describe a self-image rather than your behaviour: who you want to be, who you have been told you are, who your current job asks you to be. From the inside these are not easy to tell apart.
Results also move with your state on the day: a busy stretch, a low period, the weeks after a change of environment. Introverted codes in particular tend to come apart from the type descriptions on web tests. (That is why this site inserts two confirming questions only for introverted codes.)
Many people also behave differently at work and at home, so which one you had in mind changes the answers. Four letters from a single test are best held as a provisional marker.
Typing in socionics rests on interview and observation, not on a test. What matters is not only what someone answers but how they answer, where they hesitate, what they leave out. The test is a support, not the decision.
And normally several people type independently and then check whether they agree. If they do not agree, no conclusion is issued and the case is left open. Leaving it open is not a failure; it is the ordinary state. Refusing to settle quickly is what makes the answer accurate later.
So the four letters you got from a web test are not a substitute for a professional judgement. They are a good entrance, but nothing has been settled there.
Types are not a tool for sorting people. The moment "he is an ESTJ, that is why" finishes the explanation, you stop looking at the person. In practice the differences inside one type are often larger than the differences between types.
Do not use it for hiring, placement, evaluation or appraisal. The official MBTI guidance says the same (how this differs from other tests). Errors in typing would turn directly into how a person is treated.
Do not assign someone a type without their consent, and do not read "needs distance" as a reason to leave. It is a statement about designing the distance, not about the relationship failing.
If there are two candidate types, read both. Which one fits shows up most clearly when the two are side by side. Reading only one, you will adjust yourself to what it says.
Keep only the parts that land. The parts that do not land are information too: that is where this person is not like that. No description fits entirely.
Hold it as a hypothesis, not an explanation. When something is not working, try moving the cause from the other person's character to the structure of the fit. That is the soundest use of the sixteen relations.
In the end, ask them. A relation description works best as an opening for a conversation, not as a verdict. "Apparently this is where we grind — does that match how it feels to you?" is worth far more than getting the typing right.
None of this is a rejection of types. It says only that results are far more useful when everyone reading them shares the same assumption: that typing carries error.
Read with the assumption in view and the sixteen relations become a usable tool. Pick your type and theirs.
Pick the two types