I am trying to customize the default Magento 2 Elasticsearch 6. Right now the default Fulltext search is not giving that good result.

For Example I have product name like ADB12355 BBB, ADB3456 AAA, etc.

I am getting proper results either when adding ADB12355 BBB or ADB3456. The problem is I also want both the products if someone just search for ADB some kind of LIKE in mysql.

Also My configuration is correct and indexes are created but So far I am not able to find the correct place to customize.

  • go to marketing -> seo and search -> search synonyms, there you can add synonyms for the adb search that will include the product names returned by search. Also ensure the name attribute is enabled for search.
    – paj
    Commented Mar 17, 2020 at 21:40

3 Answers 3


Give a try.

Add plugin on the below class Builder.php by making a sample module.

File: app/code/Custom/LikeSearch/etc/di.xml

<type name="Magento\Elasticsearch\Model\Adapter\Index\Builder">
        <plugin name="custom_like_search_builder" type="Custom\LikeSearch\Plugin\Model\Adapter\Index\CustomBuilder"></plugin>

File: Custom\LikeSearch\Plugin\Model\Adapter\Index\CustomBuilder.php

namespace Custom\LikeSearch\Plugin\Model\Adapter\Index;

class CustomBuilder {

    public function afterBuild(\Magento\Elasticsearch\Model\Adapter\Index\Builder $subject, $result)
        $likeToken = $this->getLikeTokenizer();
        $result['analysis']['tokenizer'] = $likeToken;
        $result['analysis']['filter']['trigrams_filter'] = [
            'type' => 'ngram',
            'min_gram' => 3,
            'max_gram' => 3
        $result['analysis']['analyzer']['my_analyzer'] = [
            'type' => 'custom',
            'tokenizer' => 'standard',
            'filter' => [
                'lowercase', 'trigrams_filter'
        return $result;

    protected function getLikeTokenizer()
        $tokenizer = [
            'default_tokenizer' => [
                'type' => 'ngram'
        return $tokenizer;

Once done perform the php bin/magento indexer:reindex catalogsearch_fulltext

It should give results like below:

For search term, ADB gives results of below products names

ADB12355 BBB


For search term, GIT gives results of below products names



  • 1
    I can confirm changing the ngram min/max values to 3 greatly improves this sort of search. You can also consider increasing the search weight of the sku attribute to improve sku searches. Great answer @sandip !
    – paj
    Commented May 10, 2020 at 10:48
  • 1
    I used this but getting all the products in search result even if I am searching any random string "shsskjkxnxs" which doesn't even exist. Please help me to achieve this. @paj Commented Feb 3, 2021 at 11:27
  • 1
    I wrote a blog post about this issue, it provides some more information blog.gaiterjones.com/…
    – paj
    Commented Feb 3, 2021 at 13:27
  • Thanks @paj!! Great Help. Commented Feb 5, 2021 at 11:43
  • I was looking to create my first custom Elasticsearch module and tried this code into custom module but it's not working...also ran s:d:c, c:f, indexing etc but no luck. Commented Mar 20, 2022 at 10:28

I appreciate this isn't a direct answer to the question however I'd recommend using https://github.com/Smile-SA/elasticsuite. It's open source, very customisable, and has some excellent features included such as autocomplete and search optimisers.

  • As A LOT of other users, "No alive nodes found in your cluster" Commented Mar 21, 2022 at 14:03

you have to set AND operator in your elastic search json queries, Override the module-elasticsearch/SearchAdapter/Query/Builder/Match.php. file in your custom module and then The code you have to edit in the build function in line no 75

if($matchKey != 'match_phrase_prefix'){$matchQuery['operator'] = "AND";}

Also make sure you are not setting the operator in match_phrase_prefix.

After this you'll get exact matches in catalog search.

This has been tested well for magento CC 2.4.0

  • This is not working with M2.4.2 any suggestion?
    – Yogita
    Commented Aug 18, 2021 at 5:50

Not the answer you're looking for? Browse other questions tagged or ask your own question.