Skip to content
Search & AI visibility

People search for what you sell. They find everyone but you.

Two surfaces now decide whether a buyer ever hears your name: the results page, and the assistant they ask instead of opening one. We do the engineering half of being found — defects fixed in the code, structured data and machine-readable files that assistants can actually read, scoring built into your CMS so it stays fixed, and content and translation kept current month after month.

10

weighted scoring dimensions per page, inside the CMS where your team edits

RTL

Arabic, Farsi, Urdu and Hebrew laid out properly, each with its own SEO

agents.json + llms.txt

published on this site — read what we would build for you

What sits inside this

Five surfaces, one standing operation

Search visibility is not a project that finishes. Below is what the work is actually made of, and how it is run once it is in place.

01 / 06

  1. Technical and on-page SEO, scored where your team edits

    We get a monthly report full of green arrows, and the phone rings no more than last year.

    • Fix crawlability, speed and structured data in the code
    • Score every page live inside the CMS
    • Generate missing metadata in a click, site-wide

    Editors see the score before they publish, not in a PDF a month later.

    Crawlability, page speed, structured data, heading order, internal linking, metadata — the unglamorous layer that decides whether a page is even eligible to rank. We fix it in the code, then build the scoring into your CMS so it stays fixed: every page and document carries a live score across weighted dimensions — meta quality, image alt text, content depth, heading structure, internal links and more — with missing pieces generated in a click and a batch mode for the whole site. This is not a plugin we resell; it is a Studio tool we build, which is why it scores your content model rather than a generic one.

    Core Web Vitals · Structured data / JSON-LD · In-CMS per-page scoring · Internal linking · Metadata generation

  2. Being found — and quoted correctly — by AI assistants

    I asked ChatGPT about our own company. It described a competitor, then confidently invented something we don’t do.

    • Publish agents.json and llms.txt at the domain root
    • Write pages an assistant can lift answers from
    • Ask the assistants, then correct what comes back wrong

    Very few firms here do this work at all.

    Buyers increasingly ask an assistant before they open a search engine, and assistants read a different surface than Google does: a structured, machine-readable description of what your business does and who it serves, agents.json and llms.txt at your domain root, clean schema on every entity, and content written so a correct answer can be lifted straight out of it. We publish those files on this site, so you can read exactly what we would build for you before you commit to anything. Then we test it the way your buyer does — putting your customers' real questions to ChatGPT, Claude and Perplexity, and correcting what comes back wrong. That is precisely why it is worth doing now.

    agents.json · llms.txt · Entity and schema modelling · Assistant answer testing · Answer-shaped content

  3. Content that answers the question your buyer actually typed

    We publish when someone remembers to, and the posts read like they were written for a search engine.

    • Plan against the questions that carry commercial intent
    • Publish on an agreed cadence, in the brand's voice
    • Attach metadata, schema and internal links before publishing

    Review stays with you: nothing goes live that your team has not seen.

    We plan content against the terms and questions that carry commercial intent for your business — not volume for its own sake — then produce it on a cadence you set, in your voice, published straight into the CMS with metadata, schema and internal links already attached. It is built as a production line rather than a campaign, because the goal is a page for every question a buyer has, still earning citations in two years.

    Search-intent planning · Editorial cadence · Brand voice · Human review workflow

  4. Translation and locale operations

    There is an Arabic version of the site. It is six months behind the English, and nobody owns it.

    • Keep every language in step as English changes
    • Give each language its own metadata and URLs
    • Flip Arabic, Farsi, Urdu and Hebrew layouts properly

    This is the ongoing half of multilingual.

    Every new page, product and post has to be translated, given its own metadata and its own URL, checked for RTL layout, and kept in step as the English changes. We run that as a standing operation inside the CMS: translations live as structured, editable content your team can correct, each language carries its own SEO and sitemap, and Arabic, Farsi, Urdu and Hebrew flip the layout properly rather than pouring right-to-left text into a left-to-right template. Building that architecture is a website engagement; keeping it accurate, current and ranking is this one.

    Arabic RTL · Per-language metadata and URLs · Ongoing sync with English · Editable in-CMS translations

  5. Competitor and content-gap analysis

    The same three competitors sit above us on every term that matters, and we can’t say what they’re doing differently.

    • Start from the competitors actually taking the deals
    • Name the terms and pages they own instead
    • Feed the gap list into next month's writing

    Benchmarking a rival is a click rather than a project.

    We take the competitors you actually lose deals to, pull apart what they rank and get cited for, and hand the gap back as a work list rather than a spreadsheet: the terms you should own and do not, the pages they have and you do not, the entities assistants associate with them instead of you. It runs as a tab inside your CMS as well, so benchmarking a rival is a click rather than a project — and the list feeds straight into what gets written next month.

    SERP benchmarking · Assistant-answer benchmarking · Gap turned into a worklist · Runs in-CMS

  6. How it runs once it is in place

    The last SEO project ended, the deck was genuinely good, and nothing has changed since.

    • Run technical checks and ship content monthly
    • Send a short written read on what moved
    • Change the code when the fix isn't content

    Nothing waits in a queue behind a developer who does not work for us.

    Visibility is not a delivery, it is a position you hold. Competitors publish, results reshape, assistants re-read your site on their own schedule, and your own team ships pages that need checking. So we work to a monthly rhythm: technical checks, content shipped, translations kept current, assistant answers re-tested, and a short written read on what moved and what we are doing about it next. The difference from an agency retainer is that the same team can change the code when the fix is not a content fix — a template, a route, a schema, the CMS itself.

    Monthly cadence · Written reporting · Same team writes the code · No handoff between agency and dev

If traffic is not your bottleneck — visitors arrive and the site does nothing with them — visibility is the wrong first spend.

Fix the page they land on, or the follow-up that never happens, and come back to this. If you already get more enquiries than your team can answer, that is an operations problem and we will say so on the first call rather than sell you more of what is already overwhelming you. And if what you need is an ad account managed and nothing else, a paid-media specialist will do that better than we will.

Delivered work

Where this is already built and running

All case studies →

The questions people ask before they book

No, and the distinction matters. We are a technology consultancy: we work out how a business should run and then build it. Search and AI visibility is one of the things we build into what we ship. We do not sell link packages and we do not manage ad spend. What we do is the engineering half — technical defects fixed in the code, structured data, the machine-readable layer assistants read, CMS tooling that keeps a content team honest, and the content and translation operation that sits on top.

You cannot buy a place in an assistant's answer, and anyone selling you one is guessing. What you can do is make your business legible to them: a clean machine-readable description of what you do and who you serve, agents.json and llms.txt published at your domain, correct schema on every entity, and content written so a usable answer can be lifted straight from it. Then we ask the assistants what your customers ask and correct what comes back wrong. It is the same discipline as technical SEO, aimed at a reader that is not a person.

No. Anyone who does is either lucky or lying. What we commit to is the work itself: the technical defects fixed, the pages shipped, the schema and machine-readable layer correct, the translations current, the assistant answers tested — and a written account each month of what changed and what we did about it. Where something is outside our influence, we say so instead of charging for it.

Usually not. Most of this can be done on the site you already have, and we would rather fix what is there than sell you a rebuild. It breaks down only when the platform itself cannot emit what search engines and assistants need — per-page metadata, per-language URLs, real structured data, your own files at the domain root. Page-builder platforms hit that wall by design, not by accident. If that is where you are, we will show you exactly which limit you are running into before anyone mentions rebuilding.

Yes. We write real code rather than assembling a vendor's blocks, so we connect to whatever you already run instead of asking you to move into our tool. Where the platform allows the fix, we make it there. Where it does not, we tell you which part of it is the constraint and what it would take to remove it.

Properly, or not at all. Right-to-left is a layout problem before it is a translation problem: mirrored navigation, correct typography, forms and tables that read in the right direction. We treat each language as its own structured content with its own metadata and its own URL so it can rank on its own terms, and we keep it in step with the English as the site changes rather than letting it drift.

You can, and occasionally that is the right call — if you have an in-house team who will act on every line of it. But an audit is a photograph of something that keeps moving: correct on the day it was written, and out of date as soon as competitors publish, the site changes and the results reshape. We would rather find the problems and then be the ones who fix them, month after month, than hand you a document and wish you luck.

The first conversation is free and it is a real conversation, not a pitch. If there is something worth doing, we run a paid discovery: what you rank and get cited for today, where the technical work sits, what your competitors hold that you do not, and what an ongoing operation should actually cover. That produces a written recommendation and a phased quote. The discovery is yours either way — you can take it and walk.

Find out where you actually stand.

Bring the terms you want to own and the competitors who keep getting there first. We will look at your site, ask the assistants what they say about you today, and tell you plainly whether there is work worth doing here — and whether we are the right people to do it.

Free, no deck, no obligation. You leave with a straight read on your visibility whether or not anything follows.