The Shortlist Method
Most agencies sell everyone the same list. We shortlist.
Your site does not need most of what the industry sells, and the parts it does need are not the parts your competitor needs. Three filters decide what makes your shortlist. All three are published here, including the tactics we refuse to sell.
How a shortlist gets made
Three filters. What survives all three is yours.
Applied in order, to your site, against what the engines actually document. Two sites in the same industry routinely come out with different shortlists, which is the point.
01
The order
What has to be true before anything else can pay?
Four gates, worked in sequence. A failing gate invalidates everything above it, so the shortlist always starts at the lowest one that is broken, not at the item that is easiest to bill.
02
The layer
Where is the leverage on this particular site?
Compliance is the floor and every competitor can reach it. Above the floor, the shortlist favours what cannot be paraphrased away over what merely rearranges what is already there.
03
The evidence
Is it documented, or is it folklore?
We keep every approach in one library and check each against what the search engines actually publish. What is documented gets used with confidence. What is inference gets used with caution and named as inference. Nothing gets sold as certainty because it is popular.
Filter one, the order
Four gates. A failing one invalidates everything above it.
Most audits hand over a list where everything looks equally urgent, so the cheap items get done first and the blocking items sit unfixed under months of content spend. Your shortlist starts at the lowest gate that is actually broken.
G0
Access
Can a search engine reach, render and index it at all?
If it fails: Nothing downstream can work. Writing category content for a page that cannot be indexed is spend with no possible return.
G1
Comprehension
Can it tell what each page is, and which URL is the real one?
If it fails: Pages compete with themselves and get read as the wrong type.
G2
Substance
Is there anything on the page worth ranking or quoting?
If it fails: The site is indexable, understood, and still has no reason to win. This is the gate most agencies skip.
G3
Reach
Can it be found, linked and cited?
If it fails: Good pages sit orphaned and uncited.
Riskan interrupt, not a gate
Live policy exposure jumps the queue from any stage. It is the only thing that outranks the sequence.
Practicepermanent
Rules governing our own deliverables rather than your site. Applies to every engagement and is never billed as site work.
Reporting follows the same rule: we report against the gates, not the item count.
Filter two, the layer
Compliance qualifies you to compete. It never wins.
Every competitor can read the same documentation, so doing it correctly is the floor. Above the floor sit two different jobs, and which one your shortlist leans on depends on what you already have.
Foundation
commoditised
Compliance with what the search engines document. Necessary, cheap to copy, and worth nothing on its own. This is gates G0 and G1, and it is most of what the industry sells as strategy.
Substance
the moat
Being the source others cite. What cannot be paraphrased away: measurements, tests, stock reality, first-hand experience. Google's own topic authority signal is original reporting cited by other publishers, which means the route to authority is being the cited source, not building the diagram.
Organisation
the arrangement
Topical architecture, semantic and entity work. It arranges what you publish and creates nothing to arrange.
Fifteen cluster pages of commodity content is scaled content abuse. Fifteen carrying first-hand data is authority.
Same architecture, opposite outcome, and the variable is Substance.
Why this decides your order: Substance before arrangement, always, because arranging commodity content faster only reaches the wrong conclusion sooner.
Filter three, the evidence
Every approach we would consider, and what the engines actually say about it.
Not ranked, and not scored. Two approaches can both be well documented and only one of them apply to your site, which is a question about your situation rather than about the tactic. What we publish instead is the position the search engines themselves have taken, including where they have taken none.
Intent and format matching
Serving the page type the query actually wants: a comparison for a comparison query, a service page for a hiring query, a guide for a how-to.
When we reach for it
First, almost always. It is the cheapest correction available and the most common single reason a good page does not rank.
What is documented
Named directly. Google documents ranking systems and passage ranking, and serving the wrong page type is a relevance failure by its own description.
Information gain, original data
Publishing something only you can measure, test, stock or witness, with the method shown so it is checkable.
When we reach for it
When a category is saturated with pages that all say the same thing, which is most categories now.
What is documented
Google's helpful content guidance names original reporting and first-hand expertise. A Google patent describes contextual estimation of link information gain, which shows the idea was filed rather than that it ships.
Semantic and entity work
Making the brand resolve as one clear thing a machine can describe, with consistent facts across the sources an engine checks.
When we reach for it
When engines describe the brand wrongly, confuse it with another, or reach for a competitor they can describe more easily.
What is documented
The mechanism is documented and named. The practice of adding more entities to a page in the hope of moving rankings is not, and we treat that as inference.
Content pruning and deduplication
Removing or merging near-identical pages so one strong page carries the topic instead of five weak ones splitting it.
When we reach for it
When a site has grown by accretion and the same query is answered by several pages that compete with each other.
What is documented
Deduplication is a named Google system and near-identical pages demonstrably collapse into one. The recovery side, how much comes back after pruning, rests on case studies rather than documentation.
Topical architecture, pillar and cluster
Arranging what you publish so a topic reads as one coherent body rather than scattered posts.
When we reach for it
After there is substance to arrange. Architecture applied to commodity content only reaches the wrong conclusion faster.
What is documented
The engines document topical relevance. The specific pillar-and-cluster model, and the figures usually quoted for it, come from vendor analyses rather than from the engines.
Digital PR and brand mentions
Earning coverage and mentions across the wider web.
When we reach for it
Rarely, and never as a route to AI citations. Where a brand genuinely has something newsworthy, it is worth doing on its own merits.
What is documented
Google states that seeking inauthentic mentions across the web is not as helpful as it might seem. We could not find a primary study behind the usual claims made for it, including the claims we made ourselves before we checked.
Programmatic pages at scale
Generating many pages from a data source, one per variation.
When we reach for it
Only where each page carries real, distinct specifications a buyer needs. Never to add page count.
What is documented
Google states that page quantity does not raise quality, and names creating separate content for every variation, done to manipulate rankings, as a scaled content abuse policy violation.
AI crawler access and eligibility
Confirming GPTBot, OAI-SearchBot, ClaudeBot and PerplexityBot can reach and render the pages that matter, and that those pages are indexed and snippet-eligible.
When we reach for it
Before anything else. It is the lowest gate and it is broken more often than owners expect.
What is documented
Each crawler publishes its user agent and its rules. Google names indexing and snippet eligibility as the precondition for appearing in its generative AI features at all.
Library last reviewed . An entry changes whenever the engines change what they publish.
The point of showing you this
Two of these we will usually talk you out of, and both are easy to sell. We would rather lose the line item than bill you for something we cannot show you the basis for.
Want to see
your shortlist?
The free audit runs the three filters on your site and comes back with the handful of things that clear all three, in order. Written within 24 hours, no payment, no sales call.