Imagine: you weekly receive Excel files from 10 suppliers with thousands of products. Operators manually create cards in the admin panel — spending days, making mistakes in prices, creating duplicates. We solved this problem for a network of 50 online stores by building fully automatic product card creation from parsed data. Result: cards are created in minutes, errors minimized, and operators only handle atypical cases. Savings on salary were significant, and project payback was a few months. This is a key element of eCommerce automation for stores with large assortments.
What problems does automation solve?
The main difficulties in manual entry are duplicate products from different suppliers, invalid data (zero prices, giant names), and category mismatches. Let's examine each.
Duplicate products
A product may appear from several suppliers. Without automation, operators miss repetitions — identical cards with different prices appear. The duplicate detector checks SKU, GTIN, and fuzzy name match with a 90% threshold (Duplicate detection), updating the existing product.
Invalid data
The parser may output a price of "0" or a 300-character name. Allowing such data into the catalog breaks sorting. The DataValidator checks all fields against business rules: minimum price, maximum length, required fields — before recording.
Category mismatch
Each supplier has their own structure. The CategoryMatcher uses three strategies: exact mapping by supplier ID, fuzzy search by name, and on failure — keyword analysis of the product name.
Also solved are variant products (size, color) and SEO filling of cards.
How the card creation pipeline works?
Let's break down a real case: a clothing store chain with 15,000 products from 20 suppliers. Each product could have 3 sizes and 5 colors — potentially 300,000 cards. Automation paid off in 2 months. Below are the key steps.
Input data validation
class DataValidator { private array $rules = [ 'sku' => 'required|string|max:100', 'name' => 'required|string|max:500', 'price' => 'required|numeric|min:0.01', 'description' => 'nullable|string', 'images' => 'nullable|array', 'images.*' => 'nullable|url', ]; public function validate(array $data): ValidationResult { $validator = Validator::make($data, $this->rules); if ($validator->fails()) { return ValidationResult::fail($validator->errors()->toArray()); } // Additional business rules if (isset($data['price']) && $data['price'] < config('import.min_price', 1)) { return ValidationResult::fail(['price' => ['Price is suspiciously low']]); } return ValidationResult::pass($validator->validated()); } } Category matching
We determine which store category the product belongs to. If the supplier has a mapping, we use it. Otherwise, we search for categories with similar names or by keywords.
Duplicate detector
Checks for product existence by three algorithms: supplier SKU, global GTIN, and fuzzy name with brand match. On match, we update price and stock without creating duplicates.
Variant builder
Groups a flat list into a configurable product with sizes and colors. For example, from 20 rows we create 1 parent card and 20 variants with distinct SKUs.
SEO filling
Meta tags and slug are automatically generated from name and brand, eliminating manual SEO work for each card.
Why automate product card creation?
Manual creation of 1000 cards takes about 80 hours and contains 5–10% errors. The pipeline processes the same volume in 15 minutes with a near-zero error rate. 10x faster — that’s 95% time savings for operators. Besides speed, you get:
| Criteria | Manual entry | Automated pipeline |
|---|---|---|
| Time for 1000 cards | 80 hours | 15 minutes |
| Error rate | 5–10% | <0.1% |
| Scalability | Hire operators | Add a supplier in 2–3 days |
| Data consistency | Variable | Unified rules, SEO filling |
We guarantee consistency: all fields filled by uniform rules, no missing attributes. Scaling supplier integration: adding a new supplier takes 2–3 days, not weeks of recruitment.
For over 5 years we have built such pipelines, delivering 50+ eCommerce data integration projects. We guarantee code operation for 12 months after delivery.
What’s included in the work?
| Component | Description |
|---|---|
| Documentation | Field mapping schema, validation rules, normalized data structure |
| Access | Setup API permissions for the store, queue system, and monitoring access |
| Training | Manual for operators on atypical cases and import log analysis |
| Support | 1 month of post-release support: bug fixes, speed optimization |
Work process: from analysis to deployment
- Analytics — study supplier data format, identify common errors.
- Design — describe mapping schema, business rules, choose matching strategy.
- Implementation — build validator, normalizer, matcher, duplicate detector, and card creator.
- Testing — test on a sample set: valid product, duplicate, typo in name.
- Deployment — deploy queue to production, enable logging and alerts.
If you want to automate product card creation, contact us. We’ll assess the task within 1 day and propose timelines. Get a consultation or order an assessment right now.