Every link building programme is a target list plus execution. The execution is a solved problem — write well, pitch one editor, do not chase. The list is where programmes are actually won and lost, and it is the part most agencies skip by buying a database and calling it research.
A database gives you sites that accept contributions. That is a different population from publications your buyers read, and the overlap is smaller than the industry pretends.
Seven places candidates come from
No single source produces a good list. Each of these finds titles the others miss, which is the whole reason to use all of them.
Competitor page-level backlinks. The obvious one and still the most productive. Take the four URLs above you, export their referring domains, subtract yours. This finds publications that already cover the category.
Where your own customers were before they were customers. Ask five customers what they read. Not a survey — five conversations. The answers are consistently more specific and less obvious than any tool output, and they surface newsletters that no database indexes.
Journalists and analysts who cover the category. Search for the category term in news results over the last two years and collect bylines rather than domains. A named writer who has covered your space three times is a better target than a site that covered it once.
Conference and podcast circuits. Whoever speaks at the two conferences your buyers attend writes somewhere. Speaker lists are public and they are effectively a curated list of people with platforms in your category.
Round-ups and comparison pages that already rank. Search the "best X" and "X alternatives" terms. Every result is a publication making category judgements, which is exactly the kind of place a placement earns its cost.
Trade associations and their publications. Almost every vertical has one, most have a magazine or newsletter, and almost nobody pitches them because they do not appear in link databases.
Practitioner communities and their newsletters. Smaller audiences, higher concentration. In vertical SaaS these are frequently the highest-converting placements available and they rarely have a domain rating worth mentioning.
Five filters, applied in cost order
Run the cheap disqualifiers first so you never spend judgement on something a number could have killed.
Page-level organic traffic. Not the domain. The kind of page you would appear on. This removes manufactured-authority sites immediately, and it removes them at almost no cost.
Topical overlap, measured. Search the site for the three terms that define your category. If nothing comes back, it does not cover you regardless of what its section headings claim.
Publishing history. Output that predates any contribution programme. A site whose archive begins the year it started accepting guest posts is inventory wearing a publication's clothes.
A real, findable author. Named, with a byline history somewhere other than this site. Anonymous or single-site authors are the cheapest reliable signal there is.
No shared footprint. Hosting ranges, templates and outbound patterns compared across the whole candidate set rather than one at a time. This is the check that per-link vetting structurally cannot perform, and it is where clusters of related properties reveal themselves.
Reachability is a separate question
The filters tell you what a placement would be worth. They do not tell you whether it is available, and a list that ignores this is a wish list with a spreadsheet's confidence.
Reachability means three concrete things. A named editor or contributions contact exists and is current. The publication actually runs work by outside contributors, visibly, in the last six months. And somebody has had a reply from them within about a year — your own outreach history if you have one, or evidence of recent outside bylines if you do not.
Score it in three states rather than as a number. Proven: we or somebody visible has landed there recently. Plausible: outside contributors appear, contact exists, no evidence either way. Unproven: no visible route in.
Then plan the programme against proven and plausible only. Unproven titles are worth attempting when the list thins, but a plan whose arithmetic depends on them is a plan that misses.
What the finished list tells you about the programme
Count what survives, and the count decides more than the ranking does.
Under thirty reachable titles. The category is small. A programme at ten placements a month exhausts the list in a quarter and then starts buying things that failed the filters. The right answer here is lower volume, higher effort per placement, and budget moved into building something citable. We capped one engagement at six a month for exactly this reason and put the remainder into a single benchmark.
Thirty to a hundred. The normal B2B SaaS shape. Supports a sustained programme for nine to fifteen months, which is usually longer than the gap takes to close.
Over a hundred and fifty. The category is broad and so is the competition. Volume is available, per-placement cost is lower, and the differentiator moves from access to editorial quality — because everyone can reach these titles.
Maintaining it
A target list decays. Editors move, publications close, contribution policies change, and a title that was open in January is closed by September. Three habits keep it alive.
Re-check reachability on the top twenty every quarter — it is an hour and it prevents pitching into a void. Log every rejection with a reason code, so the same candidate does not get re-argued twice a year. And add new candidates continuously from the seven sources rather than rebuilding the list annually, because the sources produce different things at different times.
The list is the asset. It is the thing that survives an agency change, it is the thing worth owning outright in a contract, and it is the only part of a link building programme that gets more valuable the longer it runs.