Product Matching and Competitor Data for a Restaurant Chain: Case Study

Use Web Scraping and Competitor Data for Real-Time Pricing

Pricing managers need accurate, up-to-date pricing data to make smart real-time decisions, and we make sure they get exactly that! One of our most complex projects was helping a national fast-food chain track and standardize competitor prices across their rivals’ websites and delivery apps like Uber Eats and DoorDash.

The goal? Enable competitive pricing decisions by identifying discrepancies in product and location-level pricing across platforms, using precise price scraping, web scraping, and web crawling services.

But this wasn’t a simple scrape-and-deliver job. This project involved tens of thousands of records, inconsistent addresses, and non-standard product names across platforms. It demanded deep technical capabilities, intelligent automation, and serious human judgment.

Challenge 1: Address Normalization Across Platforms

Franchisees input their own location data on third-party apps, resulting in misalignments such as:

• Suite numbers present on one platform, missing on another

• Typos in street addresses (e.g., 123 vs. 124)

• Missing street direction (e.g., "North" vs. none)

• Incorrect GPS coordinates

  
  

With hundreds of locations and three different delivery platforms, aligning addresses required more than basic scraping.

How Ficstar Solved It

Using our proprietary web crawling services, we:

• Scraped all location data from the brand’s site and matched it against third-party platforms

• Standardized addresses through normalization rules (abbreviations, casing, syntax)

• Cross-referenced with phone numbers, zip codes, and geolocation

• Flagged potential mismatches for human review when location accuracy wasn’t 100% confident

  
  

This hybrid approach allowed us to build accurate, scalable location mapping for competitor price scraping.

Challenge 2: Product Matching With Inconsistent Naming

Unlike locations, menu items don’t have coordinates, and product names varied significantly:

• On the official site, an item might be "Crispy Chicken"

• On DoorDash, it was "Chicken Sandwich"

• Some entries included size descriptors ("Medium Chicken Sandwich"), others didn’t

• Other listings omitted key ingredients or renamed products entirely

"Ensuring consistency depends on the type of data we’re dealing with because data is always very contextual. What consistent means can vary from project to project, making it difficult to provide a one-size-fits-all answer." - Scott Vahey, Director of Technology at Ficstar

For pricing managers, this made competitor price monitoring nearly impossible without standardization.

Ficstar’s Approach

We used Natural Language Processing (NLP) to:

• Analyze word similarity, order, size descriptors, and synonyms

• Automatically match high-confidence items

• Flag edge cases for manual review

• Maintain a product-matching reference map for ongoing use

This enabled the client to receive structured, verified pricing data that accurately reflected identical products—even when naming differed.

How Does Ficstar Handle Discrepancies?

In complex data environments like this, discrepancies are inevitable. Our solution is built around a two-phase approach that combines human accuracy with machine-driven efficiency.

Phase 1: Manual Review & Confirmation

During the first pass, we manually review all ambiguous matches. While our code identifies likely issues, some competitor data is highly contextual.

Example: If a scraped item is labeled "Dryer Vent", how do we know it’s really a dryer vent?

• If it’s under a "Home Hardware > Ventilation" category, we might infer it

• If not, we investigate manually

This principle also applies to prices:

• If a price jumps 20% or more, we flag it

• If a product that was $8.99 suddenly becomes $24.99, we verify it with the client or by crawling a second time

Phase 2: Automated Monitoring & Variance Thresholds

Once the initial data is validated, we implement variance tracking:

• We set thresholds for price fluctuations, product name changes, and category mismatches

• We monitor for new entries and unexpected changes on every scheduled crawl

• If a product name changes from “Dryer Vent” to “Toilet”, we flag it

• If a price moves beyond historical trends, we investigate

This incremental model means pricing managers only review what matters—and we maintain data quality at scale.

Why This Project Was Complex

✅ Thousands of products and locations

✅ Multiple external platforms with unstructured, user-generated data

✅ Different rates, fees, and pricing models by platform

✅ Franchise-level menu customization

✅ The need for ongoing real-time pricing updates

It was a true test of the power of web scraping, price scraping, and intelligent product mapping.

Results: Real-Time Competitive Pricing Insights Delivered

With Ficstar’s custom-built web scraping solution, the client now has:

• Accurate competitor price scraping across platforms

• Validated, structured pricing data for analysis

• Real-time visibility into price variances

• Confidence in their competitive pricing strategy

• Scalable automation with human-level accuracy

This is what effective web crawling services are all about: delivering reliable, actionable pricing data that pricing managers can use immediately.