Panda Data Scraping

Ramadan is a phase and the calendar will not align it

Ramadan is the dominant demand event in Saudi grocery, and it moves roughly eleven days earlier each Gregorian year.

So a conventional year-over-year comparison is unreliable:

• The same calendar week can be mid-Ramadan one year and ordinary trading the next.
• Category demand shifts dramatically during it — dates, staples, beverages and prepared foods especially.
• Promotional intensity peaks and then shifts again for Eid.
• Trading hours change, which moves when demand lands within the day.

We deliver days_from_ramadan_start and ramadan_phase on every record so a series can be aligned to the phase rather than to the date — the approach our GCC coverage page sets out in full.

What we do not do

Apply a seasonal adjustment. Which phases matter and how to weight them is an analytical decision, and we provide the alignment field rather than a pre-adjusted series.

Some staples are regulated, and that changes what a flat price means

Saudi authorities operate price oversight on certain staple goods. Where that applies, a retailer's price is constrained rather than freely chosen.

That matters for interpretation:

• A flat price on a regulated staple is not evidence of pricing discipline or competitive positioning.
• A competitive index including regulated lines measures regulation in those lines rather than competition.
• Cross-market comparison is affected, since the same product may be regulated in one market and free in another.

Where a product is publicly identified as sitting under price oversight, we record price_regulated_flag. Where it is not publicly identified, we do not infer it — a stable price is not evidence of regulation, and assuming otherwise would put a policy judgement in a data field.

We report regulated_share_category where the flag is available, so an index can exclude or segment those lines deliberately.

Arabic listings, dayparts and what we do not collect

Arabic

Product names are retained in Arabic exactly as published, with text_direction preserved. Translation is additive and dated, never a replacement — the position our multilingual page argues.

Mixed-direction strings are common, since Latin-script brand names appear inside Arabic product names. We extract brand_string_extracted as a separate matching signal.

Dayparts

Trading patterns follow prayer times and, during Ramadan, shift substantially into evening hours. A single daily observation lands at an arbitrary point in that pattern.

daypart is derived and recorded, and we recommend multiple observations rather than one — the argument our cadence page makes about timing beating frequency.

What we do not collect

• A regulation status we inferred. Only where publicly identified.
• Loyalty account data. No accounts created.
• Sales, volumes, customer or employee data.