Skip to content

Search Result Checker

Overview

The Search Result Checker provides insights into how a search query is processed to generate a search result. It is useful for:

  • Debugging search issues.
  • Analyzing search configurations before making them live.

Running the Checker

  1. Enter a search term (word or product number).
  2. Click “Search” to execute the query.
  3. If results are found, all related search processing details will be displayed.
  4. If no results are found, an error message will be displayed.
  5. Use the buttons at the bottom to control how many search results are shown:
    • Show first 1000 results Show first 1000 results
    • Show all results Show all results

Columns in the Backoffice

ColumnDescription
ActionThe type of processing step applied to the search term.
wordListThe processed word list resulting from the action.
cutWordListA secondary word list, also generated from the action.

Special Notations

  • ”¡” (Exclamation Mark) → Represents negation (e.g., “without” or “not”).
  • Example: Searching for "Bra without wire" → In wordList, it appears as "¡wire", ensuring proper negation handling.

Processing Steps and Actions

ActionDescription
origSTermOriginal search term as passed to the EPS.
replaced special charactersRemoves special characters.
Example: "jack & jones""jack jones".
normalizedRemoves punctuation and converts uppercase letters to lowercase.
Example: "A.S.98""as 98".
killedStopwordsRemoves stopwords.
Example: "eine tür""tür".
rebuild vowelsConverts "ae" into "ä" to allow further normalization to "a".
handledReplacementsApplies replacements if configured.
cutWordList after first roundCreates word segments from the normalized string.
killed stop in cutWordListRemoves stopwords from the cut list.
stemmed wordListStems words to their root.
Example: "hosen herren""hose herr".
stemmed cutWordListStems the words in the cut list.
kill double terms in cutWordListRemoves duplicates in the cut list.
cutWordList removedRemoves words already in the main list.
wordListFinal words used in the search index, including ranking malus values.

Examples of Search Processing

Example 1: “rotes dirndl”

(rot#0.0) & (dirndl#0.0 | landhauskleid#2.0 | tracht#2.0 | trachtenkleid#2.0)

  • “rot” is stemmed and carries no malus.
  • “dirndl” includes bidirectional synonyms with a 2.0 malus.

Example 2: “fernsher” (misspelled “fernseher”)

(fernsh#0.0 | fernseh#1.556 | fernsehapparat#3.556 | fernsehgerät#3.556 | flachbildfernseher#3.556 | tv#3.556)

  • “fernsher” is corrected.
  • Spellchecker assigns a malus based on edit distance.
  • Synonyms inherit additional penalties.

Query Processing Variants

Variant 1: Query Reduction

StepDescription
Alternatives TestedLists all attempts to simplify the query.
Final Reduced QueryQuery that produced a valid result.

Variant 2: Deep Query

Used when Query Reduction fails.

StepDescription
Deep Query RequestedMarks usage of Deep Query.
Alternative Search TermsShows top 10 alternatives considered.
Final Alternative Search TermBest selected alternative.

Variant 3: Blacklist Handling

Prevents Query Reduction and Deep Query if a term is blacklisted.

StepDescription
Query Reduction not used"is blacklisted"
Deep Query not used"is blacklisted"

Variant 4: No Available Products

Deep Query is skipped if results exist but are filtered out (e.g., out of stock).

StepDescription
Deep Query not used"had results before filtering"

Replacement Processing

If replacements are applied, additional data is shown:

StepDescription
replacementExample: combinedReplacement with category restriction and replacement type.
conditionsReplacement configuration (e.g., restricted category).
found replacementName of the replacement rule used.

Search Result Information

FieldDescription
Major MKZDominant MKZ among results.
Major MKZsAll MKZs shown with product counts and percentages.
SortimentTop-level MKZ category.
CutOff LimitMinimum relevance score for results.
Found StylesTotal matched styles.
Showing StylesStyles displayed in the search preview.

Filter Raw Data

Raw data based on all search result matches.

FieldDescription
NameFilter label (technical name in parentheses).
StylesProduct count per filter.
ValuesNumber of distinct values per filter.

Displayed Filters

What is actually shown to the user in the frontend.

FieldDescription
NameFilter name (with internal ID).
ValuesAll visible values, hover shows frontend display version.

Displayed Products (Styles)

All search result products in order.

FieldDescription
ImageProduct image used in the SERP.
SKU (AKL…)Unique product SKU.
Short DescriptionProduct title.
Relevance ScoreScoring based on query match and/or replacement logic. 200 = boost from addAndPrefer.
Bonus PointsPoints added by the Business Rule Tool.
Integrated PointsCombined relevance and rule score.
CommentExplains what matched and why the product was ranked (e.g., matched brand, filter, synonym, replacement, etc.).