Starting a family history does not require an enormous tree. A small, well-documented question gives you somewhere useful to begin. Choose one person you can identify confidently and one thing you want to learn about their life.
Begin with a home inventory
Gather what you already have permission to use: photographs, letters, certificates, family notes, and the names of people who might remember more. Keep the original order of an album or bundle of letters until you have recorded it. The way items were kept together may provide context that disappears when everything is sorted into new piles.
Make a simple list rather than trying to interpret everything immediately. Note what each item is, who currently holds it, and whether you have a copy. Distinguish a document you examined from a relative's description of a document you have not yet seen.
Choose a question small enough to answer
“Find the whole family” is an aspiration. “Where was my grandmother living when she married?” is a research question. It identifies a person, an event, and a kind of information you can seek.
Starting question: Where did Rosa Bennett live at the time of her marriage? Known: A family note gives a marriage year of 1938. Unknown: The town, the exact date, and the source of the note's information. Next step: Ask who wrote the note and locate a marriage record for the relevant place.
This is a fictional example. It shows why a family note is a lead rather than automatic proof.
Record the path, including dead ends
Write down the collection searched, name spellings used, date range, and result. A search that finds nothing can prevent needless repetition, but it does not prove that an event never occurred. The collection may be incomplete or the person may be indexed differently.
The U.S. National Archives introduces several major federal record groups used in genealogy. Use its guide to identify possibilities, then check which institution actually holds the local record you need.
Finish the first session with a next step
Save copies with understandable filenames, record their sources, and write a short summary of what is known and uncertain. Stop with one next action. A modest project you can explain and resume is more useful than a large tree whose connections you cannot trace.