Search relevance · Elasticsearch

Your search returns results. Not the right ones.

I find out why and fix it. Relevance audits for engineering teams running Elasticsearch, OpenSearch or Solr, by a search engineer with 9 years of production search work.

Said BouigherdaineAudits €5k, fixed price · Author of ScaledSearch, open source
Companies I've worked with
  • L'Oréal
  • Renault
  • Orange
  • La Poste
  • LexisNexis
  • Nexity
  • Salsify
Stack I work in
  • Elasticsearch
  • OpenSearch
  • Apache Solr
  • Apache Kafka
  • Redis
  • Java
  • Python
  • Kubernetes
  • AWS
  • Google Cloud
wireless headphones
  1. #1
    Replacement ear padsAccessories
  2. #2
    USB-C charging cableCables
  3. #3
    Headphone stand, walnutDesk accessories
  4. #11
    Wireless headphones ProHeadphones · Bestseller

Before the audit: accessories rank first and the bestseller headphones rank 11th.

Example · invented
The problem

Search fails quietly. It never shows up in a dashboard.

Nothing crashes. People type, get results, and leave. These are the patterns I see most often.

01

The right item is on page two

What people want exists, but accessories and near-matches rank above it.

02

Zero results for common queries

Plurals, typos, synonyms and other languages return nothing, and nobody reviews those queries.

03

Ranking rules nobody owns

Boosts added years ago still shape every result, and no one remembers why.

04

Changes shipped without measuring

Upgrades and migrations go live without checking whether results got better or worse.

Said Bouigherdaine
Who I work with

Search engineering for teams whose revenue depends on search.

Online shops, marketplaces and content platforms, where people come to find something. When they can't find it, they don't buy, don't read, and don't come back.

I'm Said Bouigherdaine. For 9 years I've built and tuned search in production: Elasticsearch, OpenSearch and Solr, and the Java and Kafka pipelines that feed them, on platforms with millions of users.

  • Elasticsearch, OpenSearch and Solr
  • Author of ScaledSearch, open-source search toolkit
  • Your data stays private: I work under NDA
The audit

Search Relevance Audit

A fixed-scope look at how your search really behaves, and exactly what to change.

Fixed price€5k
About 1 week
What you get
  • Your top queries judged: which results are right, which are wrong
  • Zero-result and low-click queries, grouped by cause
  • A review of your analyzers, synonyms and ranking rules
  • A ranked list of fixes, easiest wins first
  • A walkthrough call with your team
Book a free 30-min search review
How it works
  1. 1
    Free 30-min callWe look at your search together. You know whether an audit is worth it before you pay anything.
  2. 2
    Fixed proposalScope and price in writing within 48 hours.
  3. 3
    The auditAbout a week, working from your query logs, index settings and live results.
  4. 4
    Report and handoverA written report, the ranked fix list, and a call to walk your team through it.
After the auditNeed hands-on help with the fixes, or a search expert on call? I offer implementation and advisory plans, scoped after the audit.
Search for AI agents

AI agents are only as good as the search behind them.

Your agent calls a search tool. When that tool returns the wrong documents, the agent answers confidently and wrongly. It's the same relevance problem, with a new kind of user.

I audit the retrieval your agents rely on: keyword, vector and hybrid search on Elasticsearch and OpenSearch.

Book a free 30-min search review
Agent trace
user"Can I return this in Germany?"
toolsearch_docs("return policy")
  1. #1Return policy, US, 2021outdated
  2. #2Shipping times FAQoff-topic
  3. #3Gift card termsoff-topic
agent"Yes, within 30 days."

The EU policy was never retrieved. The agent was fine. The search wasn't.

Example · invented, not a client
The work is public

Don't take my word for it. Read the work.

Client results stay under NDA. What I can show you is what I build and write in the open.

ScaledSearch

Open-source command-line toolkit to migrate, audit, benchmark, monitor and tune Elasticsearch and OpenSearch clusters.

npm install -g scaledsearchView on GitHub

Scaled Search newsletter

Deep dives on search relevance and running search in production, published on LinkedIn.

Subscribe on LinkedIn
FAQ

Questions before you book

What does the audit cost?
€5k, fixed price. You get the scope in writing within 48 hours of the free call, before any work starts.
Which search engines do you work with?
Elasticsearch, OpenSearch and Solr, including vector and hybrid search on Elasticsearch and OpenSearch.
Why no case studies or testimonials?
Client work is covered by NDAs, so I don't publish results. Your search stays your business.
What do you need from us?
Usually your query logs, your index settings and an hour with whoever owns search. We agree the exact access on the free call.
Where do you work, and in which languages?
Remotely, with clients in US, European and Middle East time zones. In English or French.
What happens after the audit?
If you want help shipping the fixes, I offer implementation and advisory plans, scoped once the audit is done.

Find out what your search is getting wrong.

30 minutes, free. We look at your search together, and you decide whether an audit makes sense.

Book a free 30-min search reviewPrefer to write? Send an inquiry