Search relevance audit: where your search breaks, and what to fix first

A fixed-scope diagnosis of your search or RAG retrieval, delivered in two to three weeks. You get a measured baseline, a ranked list of failures, and the cheapest fix for each, before you spend on anything bigger.

Book a free intro call What you get
Is this a fit?

A good fit if…

  • You suspect search is costing you, but can't say where or how much.
  • You're about to spend on a new engine, a bigger model or fine-tuning, and want evidence first.
  • You want an outside, senior view before deciding what to build or who to hire.

Look elsewhere if…

  • You already know the fix and need hands to build it: a contract engagement fits better.
  • You need a security or infrastructure review. This audit is about relevance and quality.
What you get

Deliverables your team keeps.

A measured baselineAn evaluation set from your real queries, with metrics you can rerun after every change.
A ranked failure listEvery failure type found, with its cause, its share of traffic, and an example query.
The cheapest fix for eachWhat to change, roughly how hard it is, and what it should be worth.
A written report and walkthroughA report your team keeps, and a call to walk through it with engineering and product.
How it runs

From brief to handover.

Week 1

Kick-off call, access to logs and your search system, and agreement on what success looks like.

Weeks 1–2

Sample and label queries, build the baseline, and trace failures through each stage of your pipeline.

Weeks 2–3

Group failures by cause, size them, and work out the cheapest fix for each.

Final

Report and walkthrough. If you want help building the fixes, that's a separate, optional engagement.

Stack

Works with what you run.

ElasticsearchOpenSearchSolrVespaAlgoliaVector databasesRAG pipelinesPython notebooks and reports
Evidence

What I've built before.

WalmartBuilt relevance for Walmart.com and the first Walmart Grocery search engine.
PinterestSearch quality tech lead.
3 U.S. patentsIn search query understanding (synonyms and query normalization), from my time at Walmart.
WritingWhy embeddings don't work for your search: the 7 failure modes I see most, with data. Read it →
Pricing and terms

How it's priced.

Search relevance auditStarting at $9,500Fixed scope, two to three weeks. One search system or RAG pipeline.
Contract engineeringStarting at $200/hourUp to 20 hours a week, hands-on in your codebase. Billed monthly.
AdvisoryStarting at $3,500/monthTwo working sessions a month, design and experiment reviews, and async questions.

Starting prices; the final quote depends on scope and follows a free intro call.

Remote, working with teams in the US, UK and Europe. Async-first, with live time during your Pacific mornings, or your morning and early afternoon if you're in Europe. Contracted and invoiced in USD through Infersense Labs LLC (Wyoming, USA).

FAQ

Questions buyers ask.

What does the starting price cover?

One search system or RAG pipeline, with logs or real queries available, over two to three weeks. Larger scopes, several systems, or human labelling at scale are quoted on the intro call.

What do you need from us?

Read access to query logs (or a sample of real questions), access to run queries against your system, and an hour or two with an engineer who knows it.

Do we have to hire you afterwards?

No. The report is written so your own team can act on it. Some teams ask me to help build the fixes; that's optional.

How is it paid?

Invoiced in USD by Infersense Labs LLC, a U.S. (Wyoming) company, half at the start and half on delivery.

Related

Also hiring for…

Free 30-minute intro call

Tell me what you're hiring for.

A free 30-minute call with me. You'll leave with a clear next step, and an honest answer on whether I'm the right person for it.

Book a free intro call Explore services