Skip to content
SearchSynth.
Content Engine

Anyone can publish more. Almost nobody publishes better.

The content problem stopped being volume years ago. Mass-produced pages get filtered by spam policy and ignored by models, and most content programmes cannot say which asset earned a single deal. We build the opposite: fewer pieces, each grounded in evidence you already own — sales calls, support tickets, your own data — passed through a gate that refuses anything thin, and measured on the decision it influenced rather than the sessions it collected.

Book a strategy callSee pricing
EVIDENCEsales callsown datasupportdraftgatepublishedmeasuredfeeds the next brief

evidence in, gate in the middle, measurement feeding the next brief

Assets a month, not twenty
2–4Assets a month, not twenty
Nothing thin reaches the index
GatedNothing thin reaches the index
Evidence you own, not competitor rewrites
First-partyEvidence you own, not competitor rewrites
Reporting against influenced pipeline
Per-assetReporting against influenced pipeline

What the engine is made of

A content service is only as good as the thing it refuses to ship. Every part of this exists either to find something true or to stop something weak from publishing.

Evidence mining

The raw material is already inside your business. We work sales-call recordings, support tickets, win/loss notes and your own product data to find the questions buyers actually ask and the answers only you can give — which is what makes a piece unrepeatable by a competitor with the same brief.

  • Sales and support mining for real buyer language
  • Win/loss analysis for the objections that decide deals
  • First-party data audit for publishable original findings

Original research & data assets

The single most citable thing you can publish is a number nobody else has. We design the study, run the collection, check the method holds, and write it up so the figures are extractable — with the methodology stated plainly enough that a sceptic can attack it.

  • Study design with a method that survives scrutiny
  • Benchmarks and index reports you can rerun annually
  • Charts and tables built for quoting, not decoration

Extraction-ready structure

Ranking is page-level; citation is passage-level. Every piece is built so individual passages answer a question completely on their own — question-shaped headings, the answer in the first sentence, short self-contained paragraphs.

  • Question-shaped headings answered immediately beneath
  • Passage-level self-containment for retrieval
  • Schema, anchors and internal links wired in at publish

The index-worthiness gate

Every draft is scored before it can publish: uniqueness against what already ranks, factual grounding, depth, and whether it says anything the first page does not. Failing drafts go back to the queue rather than to the index.

  • Uniqueness and depth scoring against the current SERP
  • Every claim traced to a source before it ships
  • A documented reason whenever something is held back

Editorial standards & voice

A written standard your team and ours both work to: what counts as evidence, how claims are sourced, what the brand never says, and the review path from draft to publish. It is the thing that keeps quality stable when volume changes.

  • House style and claim-sourcing rules, written down
  • Named reviewer and approval path per asset type
  • Disclosure norms for AI-assisted drafting

Comparison & decision content

The pages closest to a purchase: honest comparisons, alternatives, and the constrained 'best X for Y' questions buyers actually type. Including where you genuinely lose, because that is what earns the citation and the trust.

  • Comparison and alternatives pages built for extraction
  • Definitional pages that own the category vocabulary
  • Objection-handling content drawn from real sales calls

Programmatic content, safely

Where scale genuinely helps — catalogues, locations, integrations — we build the pipeline that generates from real data and blocks anything that fails the gate. Scale without the gate is how sites get filtered.

  • Templates driven by structured first-party data
  • Per-page quality thresholds before publication
  • Crawl-budget and canonical design for large sets

Distribution & off-site placement

Publishing is not distribution. We place research where your buyers and the answer engines already look — industry press, review platforms, ranked listicles, expert communities — and log what each placement changed.

  • Placement on the third-party sources models cite
  • Expert participation in the communities that matter
  • A logged before-and-after for every placement

What you get every month

  • Two to four finished assets

    Published, structured, schema'd and internally linked — not drafts handed over for someone else to finish.

  • An evidence bank

    Mined quotes, objections, statistics and findings from your own business, reusable across sales and marketing.

  • Gate reports

    What was held back and why. The refusals tell you more about the programme than the publications.

  • Editorial standard

    The written rules the work runs to, owned by you and usable after the engagement ends.

  • Original research, quarterly

    One study a quarter designed to be rerun, so it compounds into a benchmark people cite by name.

  • Per-asset performance

    Citations earned, rankings held, and influenced pipeline — reported per piece, including the ones that failed.

How we work

Evidence

  • Gong
  • Fathom
  • Zendesk
  • Intercom
  • Your warehouse

Research

  • Survey design
  • First-party analysis
  • Public datasets

Publishing

  • MDX in your repo
  • Headless CMS
  • WordPress
  • Webflow

Quality

  • SERP differential scoring
  • Claim sourcing
  • Human review

Structure

  • Schema.org
  • Internal link graphs
  • Heading anchors

Measurement

  • GA4
  • Search Console
  • Citation tracking
  • Self-reported attribution
Process

The first 90 days

  1. Weeks 1–2

    Mine the evidence

    Calls, tickets, win/loss notes and your own data, read properly. This is where the unrepeatable material comes from, and it is the step most programmes skip.

  2. Weeks 3–4

    Standard & first briefs

    The editorial standard agreed and written, the gate criteria set, and the first briefs built around questions we know buyers ask.

  3. Weeks 5–8

    Publish and structure

    First assets live, schema and internal links wired, anchors in place. Distribution begins on the sources your buyers already read.

  4. Weeks 9–12

    First research asset

    The study designed in month one publishes, with a method built to be rerun so it becomes a benchmark rather than a one-off.

  5. Ongoing

    Compound and prune

    Refresh what is decaying, expand what earned citations, and retire what did neither. A content library needs weeding as much as planting.

Content engagements

Priced per programme rather than per word, because a word count is a measure of effort and not of value.

Content Audit

$4,500

fixed fee · 2 weeks

What you already have, what is working, what is quietly hurting you, and what to write next.

  • Full inventory with performance and decay analysis
  • Thin and cannibalising content identified
  • Evidence-source audit across sales and support
  • Prioritised brief backlog with rationale
  • Editorial standard, first draft

Teams publishing steadily with nothing to show for it.

Get started
Most chosen

Content Engine

from$7,500

per month · 6-month minimum

The full programme: evidence mining, two to four gated assets a month, structure, distribution and measurement.

  • Everything in the Audit
  • 2–4 published assets per month
  • Quarterly original research asset
  • Index-worthiness gate on every draft
  • Off-site placement and distribution
  • Per-asset reporting against pipeline

Brands who need to be the source, not another rewrite of page one.

Get started

Research Programme

from$18,000

project · per study

One original study designed, run, written and distributed — the asset your category cites by name.

  • Study design and methodology review
  • Data collection and analysis
  • Report, charts and extraction-ready summary
  • Press and analyst distribution
  • Rerun playbook so it compounds annually

Categories where nobody owns the definitive number yet.

Get started

Prices are starting points, not quotes. Scope drives the number, and you get a fixed one before you commit.

Content Engine

Content Engine questions, answered

Isn't two to four pieces a month far too few?
It is far fewer than a volume agency will quote you, and that is deliberate. Publishing more thin pages than your competitors is a race that search systems and answer engines now actively penalise, and it dilutes the crawl and internal-link equity of the pages that do work. Fewer, better, and genuinely unrepeatable beats a content calendar every time.
Do you use AI to write it?
For drafting from evidence we have gathered, yes, and we will tell you so — pretending otherwise in 2026 is not credible. What we do not do is let a model generate claims. Every fact traces to a source, every draft is edited by a human who understands the subject, and the gate rejects anything that reads as generated filler. The distinction that matters is grounding, not authorship.
How is this different from the content inside your GEO retainer?
GEO includes enough content to move the queries you are losing — it is a component of a visibility programme. This is content as its own discipline: evidence mining, original research, editorial standards, distribution and pipeline measurement. Plenty of clients run GEO alone and never need this. If your problem is that your content says nothing a competitor's does not, this is the one that fixes it.
Who actually writes it?
Writers who have worked in or covered your sector, paired with the strategist who mined the evidence. We do not use a content mill, and we do not hand your account to a junior after the sale. If we cannot staff your subject properly we will say so rather than take the retainer.
How do you measure content that mostly influences rather than converts?
Three ways, and we report them separately rather than blending them into a flattering number. Citation and ranking performance per asset. Influenced pipeline, using multi-touch where your CRM supports it. And self-reported attribution — a 'how did you hear about us' field — which for content is frequently the only place the effect becomes visible at all.
What happens to our existing content?
The audit sorts it into keep, improve, consolidate and delete. Deleting is usually the most valuable and the least popular recommendation: thin pages compete with your good ones for crawl budget and dilute topical clarity. We will show the reasoning per URL rather than asking you to take it on faith.

Bring us the piece you were proudest of that did nothing.

We will tell you honestly whether it was the topic, the structure, the evidence or the distribution — and what the version that works looks like.

30-minute strategy call

With an engineer, not a closer

Book a strategy call

Typical reply time: under 4 business hours.

hello@searchsynth.ai