Weekly Circular Data: Tracking Sale Prices Across the Top 10 US Grocery Chains

Introduction

A team tracking only the online catalogue will miss most of the promotional signal in US grocery. The weekly circular is a separate publication, produced by a different team on a different calendar, and the product page frequently does not reflect what the ad is advertising.

That gap is not an edge case. It is the normal state of affairs. If your promotional intelligence comes only from product pages, you are measuring a partial picture and reporting it as complete.

Our promotional price monitoring service and grocery catalogue data collection are built specifically to run both sources together rather than treating the circular as an afterthought.

What a circular is, and why it disagrees with the site

The circular — weekly ad, flyer, insert — is the retailer's planned promotional communication for a defined date range, targeted at a defined group of stores. It is produced weeks in advance, printed or published on a fixed schedule, and it is the authoritative statement of what the retailer intends to promote.

The online catalogue is a different system. Reasons they disagree:

So the circular and the catalogue are complementary sources answering different questions. The catalogue answers "what does it cost right now, here." The circular answers "what is the retailer promoting, to whom, and until when." Promotional strategy analysis needs the second.

Circular formats and how each is extracted

Circulars appear in three broad forms. Effort and reliability differ sharply.

Always check for the structured source first. A surprising number of interactive viewers are driven by a clean offer feed, and reading it turns a document-understanding problem into a field-mapping problem.

When OCR is required, the hard part is not reading characters. It is deciding which price belongs to which product. A circular page is a dense visual layout where adjacency carries meaning. A price printed large in the middle of a block may govern six surrounding items or only one.

Practical rules for OCR-based extraction:

• Segment the page into offer blocks before reading text, using layout structure rather than reading order.
• Treat the largest text in a block as the candidate price and the nearest product description as its subject.
• Parse fine print separately — limits, conditions, and required quantities live there, and they change the effective price.
• Route low-confidence blocks to review rather than publishing them.
• Cross-validate extracted prices against the catalogue price for the same item. An extracted price far outside the plausible range is probably a misread.

OCR extraction accuracy rate by circular format, measured against manual review

Our US retail data services handle all three circular formats, with the extraction route chosen per retailer rather than assumed.

The offer schema

An offer is not a price. It is a conditional statement about a price, and the conditions are most of the value.

Two confidence scores, not one. extraction_confidence says "did we read this correctly." match_confidence says "did we link it to the right catalogue item." They fail independently, and conflating them makes debugging impossible.

The full offer record is delivered alongside catalogue prices in our pricing and product data service, with both scores preserved per record.

Promotional mechanics that need care

Percent-off and dollar-off offers cannot be resolved from the circular alone. They need a base price from the catalogue for the same store group, on the same date. This is the strongest argument for aligning the two sources rather than running them as separate reports.

Spend thresholds are basket-level. Allocating a share of a "$5 off $25" offer to individual items produces numbers that look precise and mean nothing.

Aligning circular data to catalogue data

The join has three keys, and all three must be right.

1. Product. Ad copy is abbreviated marketing text, not a catalogue title. "Coke 12pk" must resolve to a specific retailer SKU. Run it through the same matching pipeline as any cross-retailer match: exact identifier where available, then normalised brand plus size plus variant, then attribute similarity, then human review. Store the confidence.

2. Store group. The ad applies to a store group, not all stores. Maintain a mapping from store_group_id to member store IDs and keep it current. Getting this wrong attributes promotions to stores that never ran them.

3. Date. Offers are valid for a window. Joining a Tuesday catalogue observation to a Wednesday ad window produces a mismatch that looks like a data error. Always join on overlapping date ranges, not on a single date.

The product-side join is a specification-matching problem in its own right, covered in our matching ad copy to catalogue SKUs service.

Once aligned, the useful outputs appear:

promo depth distribution by category across tracked chains share of circular offers with no matching online price change

The weekly collection calendar

Most US chains flip their ad midweek, commonly Wednesday, though regional banners vary and some run Sunday or Thursday starts. Build the calendar per banner rather than assuming a single flip day.

A workable weekly rhythm:

Two operational rules matter. First, keep the previous week's ad available — overlapping windows and late corrections are common. Second, snapshot the source document. When an extracted offer is challenged, the archived page settles it.

Retailers do issue corrections mid-window. Version offer records rather than overwriting them, so history remains reconstructable.

The extraction pipeline that powers this calendar is part of our ecommerce data extraction service.

Conclusion

Actowiz Solutions collects weekly circulars across US grocery banners, extracting structured offers with mechanic type, conditions, store group, and validity window, from structured feeds, interactive viewers, and PDF-only ads alike.

Circular offers are aligned to store-level catalogue prices so promo depth and ad-only offers can be measured rather than inferred.

You can also reach us for all your mobile app scraping, data collection, web scraping, and instant data scraper service requirements!

FAQ

Why does the weekly ad price differ from the price on the website?

The ad and the online catalogue are separate systems. Many ad offers are in-store only, require a clipped digital coupon, or apply to a store group the online store is not part of. Mechanics like BOGO also cannot be expressed as a single PDP price.

What formats do grocery circulars come in?

Three. Some interactive viewers are driven by a structured offer feed, which is the most reliable source. Others are image-based flipbooks with clickable hotspots. Some chains publish only a PDF, which requires layout analysis and OCR plus validation.

How do you link a circular offer to a specific product?

Ad copy is abbreviated, so it runs through a matching pipeline: exact identifier where present, then normalised brand plus size plus variant, then attribute similarity, then human review. Each link carries a confidence score, separate from extraction confidence.