What stops the web research from returning the same sites every time?

Each run uses a different research angle from a rotating set of six, and once the full cycle is used it moves deeper into the results pages. Anything already in your library, your own domain and confirmed competitors are excluded rather than found again, and domains whose analysis failed are held back for a cooldown period before they can return.

How is the fit score calculated?

Three components with fixed weights: project fit at 45 percent, domain quality at 30 percent and placement fit at 25 percent. A language model explains each component from the evidence, and deterministic code owns the weighting and the final number, so the same evidence always produces the same score. Every score carries its reasons, its confidence and any evidence gaps.

Which metrics sit behind the score?

Domain Rating by Ahrefs, referring domains, total backlinks, a domain rank, spam signals and monthly organic and paid search traffic estimates. Domain Rating is checked first as a free gate, so a site below your floor is dropped before any paid lookup is made for it.

Does finding a prospect send an email?

Never. Discovery only fills your project library. Outreach requires two separate deliberate actions from you: adding the prospect to a campaign, then approving the first message. Pausing the agent stops scheduled follow-ups as well.

Can I use my own qualification rules?

Yes. Set a Domain Rating floor, traffic expectations, target markets and topics or industries to exclude. Those rules join your project brief in the rubric, and changing them refreshes the score on a later run instead of leaving a stale number in place.

What happens to sites that are not publishers?

Search engines, social networks, video platforms, forums, marketplaces, app stores, reference sites, generic hosted-content platforms and government or academic domains are filtered out of web research before scoring, because they are not places you can earn an editorial link.

What do I get at the end of discovery?

A project-wide prospect library where every row carries a fit score from 0 to 100, a match label, the site type, its metrics, the evidence behind the judgement and where it came from. From there you select the publishers that enter each campaign.

Discover and qualify

Find every site that could link to you.

Not only the publishers linking to your competitors. LinkIntel also searches the open web for the listicles, roundups, guides and industry publications that rank for your topics, then scores every candidate against your project before you spend one email on it.

Choose a plan See how prospects are found

The initial scan is included with every plan. Discovery never sends email.

Prospect libraryIllustrative workflow

Visibility goal

Google rankings and AI citations

3 competitor gap2 web research

Publisher typeFound byFitVerdict

Industry publicationCompetitor gap92Strong match

Tools listicleWeb research88Strong match

Specialist blogCompetitor gap74Good match

Sector newsletterWeb research71Good match

Niche review siteImported58Conditional

Sites that already link to you are removed before scoring. Nothing is sent until you add a prospect to a campaign.

Discover

Qualify

Find contact

Pitch

Negotiate

Approve

Fulfill

Verify

The limit of competitor-only research

Copying a competitor backlink profile has a ceiling.

Competitor links tell you where a publisher has already said yes, which makes them the fastest place to start. But they only describe the past, they miss every relevant site your competitors never reached, and once the obvious overlap is used the list runs out. Research that stops there stops growing.

Proven, but finite

A competitor profile is a fixed set. Work through it and there is nothing new to find until they earn more links.

Blind to fresh pages

The roundup published last month, the buyer guide now ranking first, the site accepting contributors today. None of it is in a backlink export.

Not built for AI answers

Assistants cite the comparison pages, lists and guides people actually read. Those pages need to be found on the live web, by topic.

Three ways in

Three sources fill one library.

Whichever way a publisher is found, it lands in the same project-wide library, scored on the same scale, with its evidence attached. You compare a competitor link and a page you found on the open web side by side.

Source 01

Competitor gap

Every site that links to a competitor and not to you. With two or more competitors, the sites linking to several of them come first, because a publisher that already covers your whole category is the strongest signal in the set.

Shared referring domains ranked before the wider union
Sites that already link to you are removed by exact verification
One real competitor source page is kept as placement evidence

Source 02

Open web research

Discovery is not limited to your competitors. LinkIntel searches the live web for the listicles, roundups, buyer guides, tutorials and industry publications that already rank for your topics, and treats the exact page it found as evidence.

Six research angles, rotated so repeat runs stay fresh
Platforms that are not publishers are excluded before scoring
Between 1 and 50 new scored domains per run

Source 03

Your own list

Bring the publishers you already know. Paste one domain, a full list, or upload a CSV of up to 50 sites per batch. They are scored against the same project brief as everything else, so your list and the research sit on one scale.

One domain, a pasted list, or a CSV upload
Duplicates and sites already in the library are never charged
Same brief, same weights, same score you can compare

Open web research

Search the web the way a researcher would.

Ask for between 1 and 50 new prospects. LinkIntel builds current-year queries from your niche, topics, keywords and geography, reads the live results, and keeps the exact page that surfaced each candidate as evidence, so placement fit is judged on a real relevant page instead of a homepage guess.

Ranked, not scraped

Candidates are ordered by search position, how many separate queries surfaced them, topical evidence, and whether the URL looks like a real article.

Fresh on every run

Each run takes a different research angle. After the full cycle it moves deeper into the results pages, so a repeat request does not return the list you already have.

Publishers only

Search engines, social networks, video platforms, forums, marketplaces, app stores, reference sites and government or academic domains are removed before anything is scored.

Six research angles, rotated

Listicles and best-of comparisons

Industry publications and blogs

Tool roundups and buyer guides

Tutorials, guides and case studies

Trends, statistics and research

Contributor and submission pages

What comes back

Tools listicle

Best-of list for your category

Sector newsletter

The gap, measured against your real profile.

A gap is only useful if it is true. LinkIntel checks each candidate against your own live backlink profile by exact domain and removes every site that already links to you, before it spends anything on analysis.

Shared links first

With two or more competitors, the sites linking to several of them are ranked ahead of everything else. A publisher already covering your category is the strongest signal available.

Then the wider set

The union of every competitor referring domain follows in rank order, so breadth is added without losing the priority of the shared set.

Your links removed

Candidates are verified against your own domain by exact match. A site linking to you from any page drops out, so you never pitch a publisher you already have.

Quality gate before cost

A free Domain Rating check runs first. Only domains that clear your floor go on to paid evidence, metrics and scoring.

The score

A number you can argue with.

Every prospect gets one score from 0 to 100 and a plain match label. Open any row and you see the three components, the reasons behind each one, the confidence, and anything the evaluation could not verify.

Strong match80 to 100

Good match66 to 79

Conditional match45 to 65

Weak matchBelow 45

Measured, not guessed

Domain Rating, referring domains, total backlinks, domain rank, spam signals and monthly organic and paid traffic estimates.

The site is classified

Editorial publication, tool directory, community, company site or unknown. The type is shown to you and never quietly rewrites the score.

Your rules join the rubric

Domain Rating floors, traffic expectations, target markets and excluded topics become part of the criteria. Change them and a later run refreshes the score.

Domain Rating by Ahrefs.

See the full scoring criteria

One prospect

Tools listicle

88Strong match

Project fit45% of the score

Exact topic overlap, adjacent use cases, shared audience and shared problem context. Not keyword matching.

Domain quality30% of the score

Domain Rating, referring domains, organic traffic, spam signals and overall authority for the market.

Placement fit25% of the score

The real page that could carry your link, read from competitor evidence or the page that surfaced the site.

A model explains each component from the evidence. Code owns the weighting, the final number and the match label, so the same evidence always produces the same score.

Quality gate

Discovery is a filter, not a firehose.

The point of research is to say no cheaply, so that the emails you do send are worth the publisher reading.

No email is sent by discovery

Research fills a library. Outreach starts only when you add a prospect to a campaign and approve the first message. Adding a prospect is not permission to send.

You never pitch an existing link

Competitor candidates are checked against your live backlink profile by exact domain before anything is scored, so sites that already link to you drop out of the set.

The same site is never found twice

Your own domain, confirmed competitors, archived rows and everything already in the library are excluded from every later run instead of being rescored.

Absent evidence is not a rejection

When there is no page evidence, placement fit stays neutral and the score is marked low confidence with the gap named, rather than turning missing data into a confident no.

How it runs

From a website to a qualified campaign.

Four steps, and the last one is always yours.

Connect your website

One public crawl builds an editable brief: what you sell, your topics, your market, your audience and what makes you different. You correct anything that is wrong.

Choose the outcome

Google rankings, citations in AI answers, or both. That goal, plus your own rules on Domain Rating, traffic and excluded topics, becomes the rubric every prospect is measured against.

Discover from every source

Run competitor gap analysis, ask for more prospects from the open web, or import your own list. Everything lands in one project-wide library with its evidence attached.

Qualify into a campaign

Sort by fit, read the reasoning, then choose which publishers enter a campaign with a target page, placement type and budget. Discovery never starts outreach on its own.

Autonomy with a clear boundary.

LinkIntel does the research and the repetitive work. You keep control of who gets contacted, what is committed, and every payment to a publisher.

You decide

Your website, market and visibility goal
Competitors and qualification rules
How many prospects each research run returns
Which prospects enter a campaign, and when

LinkIntel handles

Competitor gap analysis and open web research
Duplicate, platform and competitor exclusion
Metrics, evidence and explainable fit scoring
One organised library across the whole project

What research costs

You pay for domains actually scored.

Research draws on one monthly capacity meter. Duplicates, sites already in your library, candidates that fail your Domain Rating floor before analysis, and anything that fails to complete are not charged.

Way inWhat it usesBounds

Initial Link Intelligence scanIncluded with every planCapped sample from one competitor

Competitor gap analysisJob base plus 5 credits per newly scored domainUp to 100 / 200 / 500 domains by plan

Open web research5 credits per new scored domain1 to 50 results per run

Your own list or CSV5 credits per newly scored domainUp to 50 domains per batch

One capacity meter across every agent. No fee for each link, and no markup on what a publisher charges.See plans and capacity

FAQ

Questions about discovery.

How prospects are found, how they are judged, and what happens next.

No. Competitor gap analysis is one of three sources. LinkIntel also searches the open web for listicles, roundups, buyer guides, tutorials, industry publications and contributor pages that rank for your topics, and imports lists you already own. Sites with no competitor link at all are found this way every run.

Start with research

See your first qualified prospects today.

Choose a plan, connect your website, and run the included initial scan. No email leaves LinkIntel until you approve it.

Choose a plan Explore outreach