Automatic Update of Product Descriptions and Characteristics

When synchronizing products from a supplier's XML feed, descriptions and characteristics get overwritten, losing manual edits by content managers. This leads to duplicate work and errors. We implemented a system with separate data storage and an override mechanism that automatically loads data but prioritizes manual changes. This approach processes up to 100,000 products in 15 minutes—30 times faster than manual updates. Over 7 years, we have completed more than 50 catalog synchronization projects. We will evaluate your project and propose an optimal solution.

Problems We Solve

• Overwriting manual edits: If a content manager manually edited a description, automation should not overwrite it. Solution: separate fields value and supplier_value with an is_manual_override flag.
• Heterogeneous data formats: Each supplier sends characteristics in their own format. A normalizer is needed to bring everything to a unified internal schema.
• Data volume: A catalog of 100,000 products requires streaming processing to avoid memory issues.
• Lack of control: Managers cannot see what changed from the supplier and cannot accept or reject changes.

How the Managed Update System Works

Key principle: separate the data source (supplier) and the final content (what is displayed on the site), with a "manually edited" flag.

CREATE TABLE product_content ( product_id int REFERENCES products(id), source varchar(30), -- supplier_id or 'manual' field varchar(50), -- description | spec_weight | spec_color ... value text, is_manual_override boolean DEFAULT false, supplier_value text, -- last value from supplier updated_at timestamptz, PRIMARY KEY (product_id, field) ); 

On automatic update: if is_manual_override = true, update only supplier_value, not value. The content manager sees the discrepancy in the interface and decides whether to accept the supplier's change. This architecture provides flexibility confirmed by practice.

Why It's Important to Separate Data Source and Final Content?

Without separation, any automatic update overwrites manual edits. Our approach with the override mechanism preserves editor changes while the supplier always sees up-to-date data. This is critical when product items are edited by multiple people.

Sources of Descriptions

Supplier XML Feed

Most manufacturing companies provide XML with extended attributes. The PHP parser reads the file streamingly using XMLReader—this allows processing catalogs of any size without memory overhead.

class XmlDescriptionSource implements DescriptionSourceInterface { public function fetch(): iterable { $xml = new \XMLReader(); $xml->open($this->url); while ($xml->read()) { if ($xml->nodeType === \XMLReader::ELEMENT && $xml->name === 'product') { $node = new \SimpleXMLElement($xml->readOuterXml()); yield $this->parseProduct($node); } } $xml->close(); } private function parseProduct(\SimpleXMLElement $node): array { $data = [ 'sku' => (string) $node['article'], 'description' => (string) $node->description, 'attributes' => [], ]; foreach ($node->attributes->attribute as $attr) { $data['attributes'][(string) $attr['name']] = (string) $attr; } return $data; } } 

API with Partial Updates

If the supplier provides an endpoint for changes, the request returns only products where at least one specified field has changed—significantly reducing processing volume.

Job Chain for Content Update

Updating descriptions is heavier than updating prices—content is large, attributes need normalization, and override flags must be checked. Optimal scheme: a separate queue with low parallelism.

class UpdateProductDescriptionsJob implements ShouldQueue { public int $tries = 3; public int $backoff = 60; // seconds between retries public function handle( DescriptionSourceInterface $source, AttributeNormalizer $normalizer, ContentUpdater $updater, ): void { foreach ($source->fetch() as $item) { $productId = Product::where('sku', $item['sku'])->value('id'); if (!$productId) continue; $updater->updateField($productId, 'description', $item['description']); foreach ($item['attributes'] as $name => $value) { $normalized = $normalizer->normalize($name, $value); if ($normalized) { $updater->updateField($productId, $normalized['key'], $normalized['value']); } } } } } 

ContentUpdater Logic

class ContentUpdater { public function updateField(int $productId, string $field, mixed $newValue): void { $existing = ProductContent::where([ 'product_id' => $productId, 'field' => $field, ])->first(); if (!$existing) { ProductContent::create([ 'product_id' => $productId, 'field' => $field, 'value' => $newValue, 'supplier_value' => $newValue, ]); return; } $existing->supplier_value = $newValue; if (!$existing->is_manual_override) { $existing->value = $newValue; } $existing->updated_at = now(); $existing->save(); } } 

Schedule and Priorities

What Other Sources Can Be Connected?

Besides XML and API, integration is possible with CSV, JSON, Excel, and direct database access to the supplier's system. The table below compares main methods.

Discrepancy Interface in Admin Panel

If value != supplier_value AND is_manual_override = true, show a warning in the product interface: "Supplier changed the value. Current: X, new from supplier: Y. Accept?" with buttons "Accept" and "Keep".

Typical Implementation Issues

What's Included in the Work

• Designing data schema and product_content table
• Implementing parser for XML feed or supplier API
• Attribute normalizer with name and type mapping
• Queue setup for background updates
• Discrepancy interface in admin panel
• Testing and documentation
• Training content managers

Contact us to evaluate your project—we will select the optimal synchronization architecture. Order catalog automatic update implementation, and your managers will stop spending time on routine edits.