E-commerce without limitations: automation of product onboarding with complete specifications

Every online store that works with multiple suppliers faces the same challenge: how to efficiently manage thousands of products when each manufacturer provides data in a different format, often incomplete and requiring extensive processing? The standard scenario looks like this - you receive an Excel spreadsheet, XML file, or link to a PDF catalog and must transform this raw data into a high-quality product offering with complete technical attributes.

The problem escalates as the product assortment expands. While introducing a few dozen products can be done manually, onboarding 5,000 or 10,000 items becomes an organizational nightmare. Research shows that 98% of e-commerce professionals consider complete product data essential for business success. Despite this, most stores still struggle with supplementing and verifying technical data, which represents a huge operational cost and a significant barrier to business development.

The consequences of incomplete product data are severe and multifaceted:

• Lower conversion - consumers leave websites with imprecise product data, looking for more reliable information sources
• Higher return rate - incomplete product attributes lead to incorrect purchasing decisions, generating a costly wave of disappointments and returns
• Weaker search engine visibility - insufficient structural data limits positioning
• Delayed introduction of new products - time-consuming onboarding process extends the launch time of new products
• Problems with multichannel sales - marketplaces reject offers without required technical attributes

getName.ai was created as a response to this specific problem. By using advanced artificial intelligence algorithms, the tool automatically analyzes product descriptions (regardless of language and format) and extracts complete sets of technical attributes that comply with required classification standards. In practice, this means reducing onboarding time from weeks to hours, eliminating costly errors, and enabling dynamic expansion of offerings without increasing the team.

In this article, we'll look at how the actual process of introducing products to e-commerce using different methods works, where the biggest challenges appear, and how getName.ai is revolutionizing this area of online store operations.

Running an online store with a wide assortment is a constant battle with the quality and completeness of product data. This is particularly acutely felt by companies working with hundreds of suppliers, where each uses its own product description standards. Let's look at the most common problems that effectively block e-commerce development and consume disproportionately large resources.

Inconsistent data formats from suppliers

The reality of electronic commerce is that each supplier sends data in their preferred format:

• Excel spreadsheets with specific column structures
• XML files with their own tags and hierarchy
• PDF documents, often protected against copying
• Links to product pages requiring scraping
• Images of technical specification tables

Moreover, even the same supplier may change the data format between consecutive deliveries. This variety of formats forces the maintenance of extensive import mechanisms and often manual data rewriting, which generates additional costs and delays.

Language barriers and localization problems

Global trade means suppliers from different countries who provide data in their local languages:

• Chinese manufacturer provides specifications in Chinese
• German supplier uses German technical terminology
• Machine translations distort technical parameters

The traditional approach requires engaging translators with knowledge of industry vocabulary or creating extensive dictionaries for mapping attribute names and values. This is a costly process prone to errors, especially with specialized technical terminology.

Bottlenecks in attribute extraction

Even when the data is already in the system, there's a problem with extracting the right attributes:

• Need to manually transcribe parameters from long text descriptions
• Extracting values from unstructured text fragments
• Converting units (e.g., inches to centimeters)
• Mapping different names of the same attribute (e.g., "Color", "Hue", "Shade")
• Standardizing dictionary values (e.g., "black", "blackness", "matte black", "RAL 9005"

One specialist can process an average of 200 products per day, which creates a serious performance limitation when dealing with thousands of new items.

Scaling problems with large catalogs

As the assortment grows, problems with data management grow exponentially:

• 10,000 products potentially mean millions of individual attribute values to manage
• Updating parameters for the entire catalog can take weeks
• Adding a new required attribute requires mass updates
• Seasonal expansions of offers generate spikes in workload for the product team

These challenges cause many stores to abandon ambitious development plans, limiting themselves not so much to a smaller assortment but to product presentations based solely on basic data. Technical specifications or product properties often remain in the form of a static table, without the possibility of using individual attributes in search filters, dynamically generated names, or advanced SEO mechanisms.

Costs of employing dedicated personnel

Attempting to solve the above problems with traditional methods leads to constant team expansion:

• Data import specialists
• Product experts familiar with industry specifics
• Programmers creating import and mapping scripts
• Data quality controllers

This generates significant personnel costs while not eliminating problems with data quality. Assuming that one specialist can process about 200 products per day, the full onboarding of 1000 products with a complete set of attributes generates a cost of at least 1500 euros - and that's just for the basic scope of work, without considering additional corrections, fixes, and optimizations.

These daily challenges not only increase operational costs but, above all, limit development potential. E-commerce stores that cannot efficiently manage a growing product catalog lose competitive advantage and expansion opportunities. In the following parts of the article, we'll show how modern AI technologies can radically change this picture.

To cope with product onboarding challenges, e-commerce stores have developed various approaches. Let's look at the most commonly used methods and their limitations, which often prevent full efficiency.

Manual data entry and scaling problems

The simplest method is manually copying data from supplier documents into the store system, often by a dedicated team:

Advantages:

• No implementation costs for advanced tools
• Good results with a small number of products (up to a few dozen)
• Ongoing quality control

Limitations:

• Extremely low efficiency (approx. 200 products per day per person)
• High cost with large product volumes
• High susceptibility to human errors (fatigue, inattention)
• Inability to quickly respond to seasonal supply spikes

Companies trying to solve the efficiency problem by simply increasing the team quickly discover that this generates new challenges related to work coordination, data consistency, and quality management.

Supplier portals and why they often fail

An alternative approach is creating a dedicated portal where suppliers themselves enter their product data:

Advantages:

• Transferring work to the supplier side
• Eliminating the data transfer stage
• Unified data format at the input

Limitations:

• Low adoption rate – suppliers typically avoid using external portals and strongly prefer transferring data in their own format, developed over years
• Data entry has a low priority for suppliers
• Need for regular training and support for suppliers
• Frequent staff rotations at suppliers require constant re-implementation
• Lack of motivation for suppliers to care about data quality and completeness

Supplier portals only work in environments where the store has sufficient negotiating power to force the use of the system. In practice, especially for medium-sized e-commerce businesses, they remain mostly unused tools.

Product classification standards (ETIM, ECLASS, GPC)

Many industries have developed unified standards for product classification and description, such as ETIM (for electrotechnical), ECLASS (for industry), or GPC from GS1 (for retail):

Advantages:

• Precisely defined data structure and attributes
• International compatibility and industry recognition
• Elimination of ambiguities and discrepancies in technical descriptions
• Facilitated data exchange between supply chain participants

Limitations:

• Long and very costly process of implementing the standard at the manufacturer
• A common standard is always a compromise - manufacturers often emphasize that it doesn't capture all aspects and unique advantages of their products
• The dynamic market development forces frequent standard updates - adapting to successive versions of classifications generates high costs for both manufacturers and recipients
• Although ETIM and ECLASS are well received in their industries, a significant number of companies still don't use these standards
• Classifications are tailored to specific industries, but there are product categories poorly described or completely omitted in these standards
• Expanding a company's assortment beyond the standard means the need to introduce alternative import methods, even if there is a well-developed process based on the BMEcat format for some products

Classification standards are a great solution that eliminates many fundamental problems with product data exchange. They introduce unambiguity, precision, and consistency at the semantic and structural levels. However, they won't become a perfect remedy until the classifications themselves are complete for all industries and until all supply chain participants (manufacturers, distributors, sellers) commonly use them. E-commerce operating across multiple industries must currently combine the benefits of standards where they are available with alternative methods for other areas.

ETL tools and attribute mapping-based systems

Technically advanced companies often implement solutions that automate the product data transformation process. Such tools can be divided into several main categories:

ETL (Extract, Transform, Load)
ETL systems specialize in retrieving data from various sources, transforming it according to defined rules, and loading it into target systems. Examples of popular tools:

• Talend - comprehensive open-source data integration platform with advanced ETL functions
• Pentaho - set of data integration and business intelligence tools enabling complex transformation flows

PIM System (Product Information Management)
PIM systems often have built-in attribute mapping mechanisms, allowing transformation of data from supplier format to system format:

• Akeneo Onboarder - solution dedicated to supplier collaboration and product data centralization
• Informatica MDM - nterprise platform for managing master data, including product data
• Contentserv - xtensive PIM system with advanced automation capabilities
• Salsify - PIM platform designed for multichannel commerce and supplier collaboration
• Pimcore – flexible open-source platform enabling advanced attribute mapping during data import. With the DataHub module, you can configure integration schemas

Dedicated product integration systems

• Productsup Supplier Onboarding - specialized solution dedicated to efficient product onboarding from suppliers

The key functionality of these systems is the ability to import data from multiple suppliers and transform it into one consistent internal classification. In practice, this means:

• Simultaneously managing dozens or hundreds of data sources (each with its own format)
• Mapping different attribute names to one target model (e.g., "color", "colour", "hue" → "Color")
• Normalizing values and units (e.g., "1500W", "1.5 kW", "1.5 kilowatts" → "1500 W")
• Transforming dictionary values (e.g., "black", "noir", "negro", "czarny" → "Black")

However, this flexibility comes at a price - each new format from a new supplier requires creating individual mapping rules.

Advantages:

• Automated process for repetitive imports
• Higher efficiency than manual entry (one specialist can manage data for thousands of products)
• Ability to handle different input formats
• Central management of data transformation rules
• Possibility of data validation during import
• Unified data model at the output, regardless of source

Limitations:

• Need to create and maintain mapping rules for each supplier and each format
• Time-consuming preparation of mapping when onboarding a new supplier (usually several days of specialist work)
• High costs of implementation and maintenance of complex systems (licenses often from tens of thousands of euros annually)
• Difficulties with handling unstructured text data and extracting valuable attributes from them
• Requires highly qualified specialists with both technical and domain knowledge
• Rigid "if-then" rules don't cope well with variants of writing the same information
• Scaling problem with very diverse assortments (hundreds of categories)
• Changes in source data structure (e.g., format update by supplier) require redesigning mapping rules

Rule-based mapping solutions work well for regular imports from the same suppliers where the data format is stable and repeatable. Their main advantage is predictability and process transparency. However, onboarding new suppliers each time requires significant work to prepare mapping rules, which extends the time to market and increases operational costs.

Moreover, traditional ETL and PIM tools often don't cope well with extracting valuable information from unstructured text descriptions, leading to incomplete utilization of available product data.

Challenges related to multilingual data

An additional complication for all previously described methods is the language barrier - when products come from international suppliers, data comes in different languages. Regardless of the chosen onboarding method, it's necessary to include a stage of translation or adaptation of data to the target language.

Although modern AI technologies have significantly improved this process and reduced its costs, it still remains an additional step in the process, generating delays and requiring additional resources. Particularly problematic is the precise translation of specialized technical terminology, which can lead to inconsistencies or errors in the final product data.

The traditional product onboarding methods presented earlier, despite their gradual evolution, still rely on the same fundamental assumptions - manual work, mapping rules, and extensive configurations. getName.ai represents a completely new approach that, instead of improving old processes, completely changes the way we think about extracting product attributes.

Artificial intelligence instead of mapping rules

Unlike traditional ETL and PIM systems that require creating complicated mapping rules for each data format, getName.ai uses advanced language models that "understand" the context and semantics of product information.

How it works in practice:

• The system analyzes all available product information (descriptions, names, specification tables)
• Recognizes and interprets the meaning of text, not just its structure or format
• Identifies key attributes and their values, regardless of how they're written
• Automatically converts and normalizes values to the required format

This fundamental difference in approach completely eliminates the need to create and maintain complex mapping rules, which are the biggest bottleneck of traditional methods.

Universality of input formats

getName.ai doesn't require any particular input data format. The system can work with:

• Any file formats (XML, JSON, CSV, Excel, PDF)
• Unstructured text descriptions (even long marketing descriptions)
• Data retrieved directly from websites
• Specification tables in any layout

This flexibility means there's no need for pre-processing or adapting data received from suppliers - it can be passed to the system in the form it was delivered.

Language independence

One of the most revolutionary features of getName.ai is complete independence from the source language of the data. The system understands the context and meaning of product attributes regardless of whether they are written in Polish, English, German, or even Chinese.

Practical aspects:

• No need to translate data before processing
• Ability to work with suppliers from around the world without language barriers
• Consistent results regardless of source language

In practice, this means that the same attribute (e.g., engine power) will be correctly recognized regardless of whether it appears in the original description as "power", "Leistung", "puissance", or "功率".

Transformation to your own product classification

The most valuable function of getName.ai is the ability to automatically transform any input data to the standardized, internal store classification. The system not only recognizes attributes but directly assigns them to a specific data model required by your business - without any mapping.

How this process looks:

• Initially, you provide a model of your own product classification: classes, attributes, dictionary values
• The system analyzes raw data from the supplier
• getName.ai identifies all available attribute information
• Automatically assigns recognized attributes to your internal classification
• Normalizes values according to requirements (units, formats, dictionaries)
• Returns a complete set of attributes ready for import into your system

This process completely eliminates the need for mapping or creating transformation rules for each new supplier - getName.ai automatically adapts to different input data formats, always maintaining the same, consistent output data format.

Intelligent unit conversion and value normalization

The system can not only recognize numerical values but also automatically identify units of measurement and convert them to the required format:

• Automatic detection of units (inches, mm, cm, kg, lb, W, kW, etc.)
• Conversion between metric and imperial systems
• Standardization of notation according to requirements (e.g., always "cm" instead of "centimeters")

The value normalization mechanism works similarly, recognizing different variants of writing the same value:

• "black", "noir", "schwarz", "czarny" → "Black"
• "wireless", "cordless", "bezprzewodowy" → "Wireless"
• "stainless steel", "inox", "stal nierdzewna" → "Stainless steel"

Quick API implementation

Integrating getName.ai with existing systems is simple thanks to the REST API interface. Let's assume our e-commerce has strategically adopted ETIM v10 classification as its main classification and wants to adapt the product description received from the manufacturer to it.

This API structure allows for easy integration with existing PIM, e-commerce, or ERP systems through standard communication mechanisms.

Scalability and flexibility

Unlike traditional solutions, getName.ai maintains high efficiency with a large variety of products or suppliers. The system handles equally well:

• One supplier and one product category
• Hundreds of suppliers and thousands of different product categories

This scalability stems from the fundamental difference in approach - instead of requiring tedious creation and maintenance of thousands of mapping rules, getName.ai uses artificial intelligence capabilities to understand the context and meaning of data, regardless of its format or source.

getName.ai is not just another improvement of existing processes, but a complete paradigm shift in the approach to product onboarding. Instead of adapting to the infinite variety of supplier data formats through difficult-to-maintain mapping rules, the system focuses on directly understanding the meaning of data and automatically transforming it into a coherent, internal classification. This is a fundamental difference that allows breaking the limitations of traditional methods and opening new possibilities for assortment development in e-commerce.

In this section, we'll look step by step at how the actual process of introducing large sets of products into e-commerce systems works. We'll show the pain points and biggest challenges faced by product teams, and exactly where getName.ai introduces significant improvements.

Real example of the onboarding process

Let's look at what a typical product onboarding process looks like using the example of an XML file from a supplier, which almost every e-commerce store must deal with:

Input data:

• XML file in the supplier's own format
• Technical specification attributes recorded in the manufacturer's own format
• Data often in a foreign language (e.g., English, German, or even Chinese)
• Catalog of 5,000 products

Etap 1: File structure analysis

This stage is common regardless of the method. The technical team must first understand how the supplier's XML file is built. This means mapping fields, identifying tags, and understanding the file logic. This stage often requires IT specialists and can take from several hours to several days, depending on the complexity of the structure.

Stage 2: Preliminary data processing

Also a common stage for all methods, including:

• Transformation to a structure acceptable by PIM/E-commerce systems
• Preparation of basic and logistical data for import
• Organization of assets (images, documentation)
• Placing specification attributes in a temporary, working structure

This stage usually requires writing conversion scripts or using ETL tools, which takes 1-2 days of IT specialist work.

Stage 3: Import of basic data

Stage common to all methods:

• Loading product identifiers
• Import of names and descriptions
• Introduction of logistical data (weight, dimensions, packaging)
• Attaching assets (images, PDF files)

Even if the data is processed automatically, this process requires supervision and takes several hours to a full day.

Stage 4: Language localization process

In this step, we need to adapt foreign-language product data to the local language of the store. This is one of the most labor-intensive stages of traditional onboarding, which getName.ai radically simplifies.

Without getName.ai:
Traditional localization requires comprehensive translation of all product data elements, which generates significant costs and delays the introduction of products to the market.

• Translation of product names
• Translation of product descriptions
• Translation of technical specifications

This process can take from several days to weeks, depending on the number and specificity of technical attributes and available translation resources.

With getName.ai:
With getName.ai, there are two main strategies: you can translate the original names and descriptions developed by the manufacturer, or focus on obtaining multilingual technical attributes and then automatically generate local names and descriptions from them.

• Translation of product names - unnecessary if the company has a mechanism for generating names based on specification attributes
• Translation of product descriptions - unnecessary if the company has an AI mechanism for generating descriptions based on specification attributes
• Translation of technical specifications - completely unnecessary, as getName.ai recognizes attributes regardless of the source language

This is where the first significant difference appears: getName.ai eliminates the need to translate technical specifications, which saves days of work and significantly reduces costs.

Stage 5: Product categorization

Each product must be assigned to the appropriate category in the system. This process usually requires specialized product knowledge and - depending on the degree of assortment diversity - can take from several hours to even two days for a batch of about 5,000 products.

Stage 6: Import of category-dependent specification attributes

Manual editing:

• Manual completion through GUI based on source file data
• Reviewing technical datasheet files for information
• Searching for missing data on the internet
• Manual unit conversion (e.g., inches to centimeters)

Assuming that one specialist can process about 200 products per day, serving 1000 products requires 5 person-days of work.

Dedicated tools / ETL:

• Preliminary data cleaning
• Creating complicated import rules [if-then-else]
• Defining mapping for each attribute
• Creating dictionary value mapping, in special cases additionally per attribute
• Configuration of automatic unit conversion

Preparing such a system usually takes 3-5 days, but after configuration, the import proceeds much faster. However, each new supplier requires reconfiguration. Moreover, each subsequent import even from the same supplier requires inspection and often modification of mapping, as new attributes, new dictionary values, or errors in source data may appear.

This solution is particularly effective for companies with a relatively simple assortment structure – mapping attributes and dictionary values remains a process that can be effectively carried out without excessive complexity. However, for e-commerce with a very diverse offer (hundreds of product categories), extensive mapping becomes practically unfeasible or requires a dedicated team for continuous rule updates.

API getName.ai:

• Full process automation
• Sending all available product information to the API without preliminary analysis
• Immediate receipt of attributes and their values ready for import

This is where the most important difference appears: a stage that traditionally took 5 person-days of work (assuming efficiency of 200 products per day) or 3-5 days to prepare advanced ETL rules, is reduced to several hours of system work. There is no need to create complicated mapping rules, and the system automatically recognizes attributes and converts units.

Stage 7: Quality control

Manual editing:

• Quality control is performed simultaneously with data entry by operators
• Does not require an additional verification stage, but is susceptible to human errors resulting from fatigue during monotonous work

Dedicated tools / ETL:

• Requires random verification, usually about 1% of products
• Errors still occur due to imperfections in mapping rules, especially with new values

API getName.ai:

• Recommended quality control covering 1-5% of products to confirm correctness
• Verification proceeds smoothly and takes several hours

Most important improvements with getName.ai:

• Elimination of language barrier - the system recognizes attributes regardless of source language
• No need to write mapping rules - the system automatically recognizes and maps attributes
• Automatic unit conversion - the system detects units of measurement and converts them to the required format
• Recognition of dictionary values - the system maps different variants of writing the same value to a standard value
• 90% reduction in time needed for technical specification stage - the most time-consuming stage is almost completely automated

Implementing getName.ai allows e-commerce companies to focus resources on strategic development of their offering instead of manual data entry, while improving the quality of product information and reducing the time to market for new products by about 80% - 90%.

Onboarding new suppliers with complete product attributes is one of the biggest challenges of modern e-commerce. In this article, we've shown how traditional approaches – despite continuous improvements – still rely on tedious, costly, and inflexible mechanisms.

We've highlighted the fundamental problems:

• Limited efficiency of manual data entry
• Language barrier with international suppliers
• Difficulties related to creating and maintaining mapping rules
• Limitations of classification standards in heterogeneous assortment

At the same time, we've shown how getName.ai – using the latest achievements in artificial intelligence – introduces a completely new quality to the onboarding process. Instead of improving old practices, it completely changes the rules of the game, eliminating the most time-consuming and costly stages.

Real benefits of implementation

As shown by analyses of processes in e-commerce companies, implementing getName.ai translates into concrete, measurable benefits:

• Drastically shortened product introduction time – from weeks to hours
• Reduction of operational costs – lower expenditure on personnel and process handling
• Elimination of language barriers – possibility of direct cooperation with suppliers from around the world
• Improved data quality – consistent, complete attributes for the entire assortment
• Unleashing team potential – specialists can focus on strategic development of the offering instead of mechanical data entry

Moreover, the specific operation of getName.ai means that benefits grow with the scale of operations. The greater the variety of products and suppliers, the greater the advantage over traditional solutions.

Start the transformation today

If you're struggling with any of the challenges described in the article, we invite you to directly familiarize yourself with the capabilities of the getName.ai platform. Free your e-commerce from the limitations of traditional methods and discover new possibilities for assortment development without compromises.

Schedule a short, 15-minute presentation during which we'll show how getName.ai handles real product data. You'll see live how raw descriptions are transformed into structured attributes - ready to use in your PIM or e-commerce system.

Test the solution without obligation

Get access to a fully functional demo version and check how the system works on a larger set of your own data. No upfront fees and no long-term commitments - see how getName.ai can support your business processes in practice.

Contact us

To schedule a demonstration or gain access to the demo environment, contact us by email hello@getname.ai or fill out the form available on our website.