Content search with NG – how do I build the optimal data feed for it?
Just like Google search, the search function in the online shop also fulfills a variety of tasks. Customers often expect not only to find suitable products - but also inspiration and advice. Be it in the form of blog posts, topic pages, videos or other content. NG can display the most relevant content for each search query and intelligently improve its sorting. But how should the data feed for the content search be best structured? Which fields are essential? And which are not? You will find out all of this in this article.
Why do I need a data feed?
The data feed is the pool of information that NG compares with the search term entered. The search results then display the data sets that best match the search term. Conversely, this means that anything that is not stored in the data feed cannot be found. This applies to products as well as to all other content.
During setup and updating, NG will import your data feed. To do this, your data must be in an organized structure. A CSV file is intended for this purpose.
How should the CSV file for content search be structured?
In order for your content to be reliably found, your feed must of course have a different field structure than a product data feed. Below are some recommendations and general tips on how to fill in the fields correctly. First of all: the text content of posts, articles, etc. should not be in the data feed. Ideally, the data feed for your search only contains as much data as necessary. Importing the complete content of blog posts or advice articles is not recommended. The large number of - often irrelevant - keywords in them tends to make the search results worse. It is much better to just include the headings in a separate column.
Relevant fields for your content data feed:
Must haves:
URL : NG must be able to display the corresponding URL for each of your content pieces. Otherwise, the search results can be viewed but not clicked on.
Image URL: This ensures that preview images of your content appear in the search results. Similar to the product name, this field contains the name of the respective content piece. The title is therefore the text that will be visible in the search results.
Short text: This text succinctly summarizes the overall content and therefore provides important keywords for the search.
Categories/headings: These allow you to implement filters for the content results in your shop. If there are many results, this can be a great help for customers.
SEO texts: SEO titles and descriptions are optimized for Google and therefore contain the most important keywords by definition. These are also relevant for the shop's internal search.
Tags/Keywords: Sometimes customers use different or colloquial terms when searching. It is therefore important to tag your content. You can see which terms your customers use in NG Analytics.
Nice to haves:
Headings: As already mentioned, the (sub)headings of an editorial article can contain search-relevant keywords.
Rating: The (star) ratings for your posts can also be displayed in the search results. This doesn't help with findability, but it is a useful ranking criterion.
Publication date: The same applies here as for the rating. It can be useful to sort the content results chronologically.
Content type: blog, video, topic page or magazine: Some customers are looking for a specific type of content. It is therefore worth maintaining this information in the data feed and implementing filters based on content type. Example: This is what an excerpt from a data feed could look like:
Other general best practices:
When filling the data feed, there are a few points you should consider:
Separate the individual fields with a separator. We use the semicolon as standard. Alternatively, you can use any other character. The important thing is to always use the same field separator for the entire file.
Always use enclosing characters at the beginning and end of each field – for example, quotation marks.
Make absolutely sure that none of the separators used appear in the article data itself. Otherwise, the field contents may become confused.
Always create the file with UTF-8 encoding.
To save bandwidth during transfer, you can also provide the file in a ZIP archive.
To use NG for both product and content searches, you should keep two different data feeds, each in a separate channel. To make it easier to assign the files to the channels, it is important to include the channel name in the file name (e.g. ng_export_de_content.csv). The files can all be stored in the same archive or individually.
Create each data record in a new line. The data in a data record must not contain any line breaks. If there are any, please feel free to bring this up with us in the project.
Each row must contain the same number of fields, regardless of whether the field is filled or not.
Conclusion: The better the data, the more reliable the content search.
Even the best search technology can only find what is stored in the search index. The quality of your data determines how quickly and easily your customers can find the right content. If you have any questions about the best practices presented, please write to us at https://www.Fact-Finder.de/kontakt.html
Would you like to learn more about content search with NG and are you interested in concrete application examples? Then I can recommend the following article on our NG blog: https://www.Fact-Finder.com/blog/ecommerce-content-search/