Before I open a single record, I start with family lore, old letters, photographs with names pencilled on the back, a Bible with births recorded in the front cover. This material is often wrong in the details — a birthplace misremembered, a date shifted by a decade, a name softened or anglicised somewhere along the way — but it's rarely wrong about the shape of the story. It's a starting point, not a source. Everything in it typically needs verifying against original documents before I treat it as fact.

From there, I work backwards. Not forwards from some distant, hoped-for ancestor, but backwards from myself: parents, then grandparents, then great-grandparents, one confirmed generation at a time. It's a slower approach than jumping straight to an appealing name several centuries back, but it's the only one that holds up. A single unverified leap — one wrong parent, one wrong parish, one assumption taken as fact — can send an entire branch down the wrong line, and everything built on top of it afterwards inherits the mistake.

Major platforms — Ancestry, FamilySearch, Find My Past — are genuinely excellent for anyone starting out, particularly for locating original documents you might otherwise spend weeks tracking down in person. What I'd caution against is accepting other users' existing research on these sites as fact simply because it's there. A tree with a thousand followers is not the same thing as a tree with a thousand verified sources. Use these platforms to search for the documents themselves, then draw your own conclusions from what's actually on the page.

Applying the work-backwards method: what I keep coming back to

Assume nothing, verify everything. Family lore is retold, and retelling changes things — sometimes by accident, sometimes because a slightly better story was more fun to pass down than the plain truth.

Document your sources as you go. It costs a few extra minutes at the time and saves hours later, when you're trying to work out why you once believed something and can no longer find where it came from.

Spelling doesn't count. Names were recorded by clerks who spelled phonetically, by immigration officials unfamiliar with the language, by ancestors who couldn't read or write themselves. A surname that shifts from Smythe to Smith to Smyth across three records typically isn't three different families — it's one family, recorded by three different hands.

Common surnames need more than a name to confirm. Where the surname is a common one, the name alone won't tell you whether you've found the right person. Place of residence, occupation, age, and family relationships are what actually confirm identity — look for the combination, not just the match.

Don't force a conclusion to prove descent from someone famous. Plenty of 19th-century family history books did exactly this, often because a claimed connection to royalty or a celebrated name sold more copies. That habit left a lot of confidently wrong trees still circulating today. Follow the facts wherever they lead, even when they lead somewhere ordinary.

Good luck with your research — and if a line stalls, that's usually a sign to go back one generation rather than jump ahead and hope the gap fills itself in. When a set of parents proves elusive, I've often found the answer by researching all of their children, not just the one in my direct line. A sibling's marriage record, obituary, or naturalization file will sometimes name a birthplace or a maiden name that never turned up in the direct ancestor's own records — the information was there all along, just filed under a different name.

If you'd rather start with a foundation already in place than begin from scratch, that's exactly what my Four-Generation Family Tree service gives you — a verified base to work backwards from, built the way I've described above.