01SEOFood and drink
Ba Li Basar
balibasar.co.il
A Hebrew knowledge site about meat, built from the ground up to be the accurate answer on Google for questions about cuts, numbers, temperatures and recipes.
Opened to Google: 378 verified pages, 106 conflicts settled
The challenge
There was no single Hebrew site that covered meat properly: cuts by the Israeli numbering system, doneness, safety, quantities and recipes that actually work. A site like that is judged on one thing, accuracy. One wrong temperature or one misnamed cut costs the reader's trust, and Google's too, since Google looks hard at content produced in volume.
What I did
- Information architecture before a single line of content. Departments for cuts by animal, recipes, cooking methods, buying, safety, a glossary, and a "behind the counter" section that explains the science. Every page type has a fixed template, and every internal link points only to a page that exists or is already on the site map.
- Binding writing rules, updated after every error found. Every professional fact rests on a source that was read in full; a search result snippet does not count as a source. Every recipe is a rewrite of one real recipe from a leading cooking site abroad: quantities, times and temperatures come from that source, with no averaging across sources and nothing invented. When that rule was set, recipes that had already been written were rewritten to meet it.
- A fact checking gate. Each page is checked against its own sources, claim by claim, and only a page that passes is marked verified. After the gate I ran a consistency round across pages: temperatures and resting times, cut numbers, and quantities against the calculators. It surfaced 106 contradictions, each one got a ruling and was checked again.
- Ten calculators (meat per person, time and temperature, thawing, dry brining, storage and more), all pulling from the same tables shown on the pages. If a table is missing, the build stops. A calculator cannot contradict the page it sits on.
- SEO shaped by search intent. I chose the long tail of cut and number questions as the battlefield, not head recipe terms owned by huge sites. Titles open with the phrase people search, descriptions answer directly, subheadings are written as questions, with FAQ schema and a printable temperature chart. Internal links are added only on phrases that already exist in the text, without rewriting copy that has already passed the check.
The decision that made the difference
Accuracy before volume. I decided Google would only see pages that had passed the fact check, and everything else would stay out of the index until it did. The site opened to Google only once all 378 of its pages were verified. Since launch, every expansion goes live in measured waves, with measurement between waves and a cap on new pages per wave. The reason: a new site that publishes hundreds of pages at once looks like mass produced content, which is exactly what Google's policy targets. A site built to be a data source has to look like one from day one.
Outcome
- Live and open to Google since October 12, 2026, two days after work began, with 378 pages that each passed the gate.
- A new spices and marinades department was built and shipped in three waves: 28 new pages, each through the same gate.
- In the consistency round, 100 of 106 contradictions were resolved, 3 were rejected with reasoning, and 3 were closed in code.
- The visible text carries no work markers ("under review", "to verify"). A claim without a source that was actually read simply does not go in, and the source list sits at the end of every page.