Red flags in a link building proposal

By Bill Hartzer · Hartzer Consulting · 25+ years in search, domains and expert witness work

The benchmark to measure a proposal against

Most proposal claims are impossible to evaluate because nobody has a baseline. Buyers have no idea what a normal rate of link accumulation looks like, so a promise of fifty strong links a quarter sounds ambitious rather than absurd.

So here is a baseline, from my own site, measured in Majestic in August 2026. It is a twenty-year-old profile in a competitive niche. It has 1,301,839 links from 22,260 referring domains. Not one of those links was ever bought.

Of those 22,260 referring domains:

• 446 — two percent — sit at Trust Flow 41 or above. Trust Flow is Majestic's measure of how trustworthy the sites linking to a page are. That works out to about twenty-two strong referring domains a year, across two decades of constant publishing, conference speaking, industry commentary and press coverage.
• 69.3% are Trust Flow 0. On a profile where nothing was purchased. Weak links are what the web produces naturally; they are not a sign that somebody cut corners.
• Five links come from .edu domains. Zero come from .gov domains. In twenty years.
• The top 100 referring domains account for 84.3% of all links, and a single domain sent 219,159 of them — 16.8% of the entire profile — because of one sitewide link.

Read a proposal against those numbers. That is the whole method of this page. When somebody offers you an accumulation rate that a twenty-year unbought profile never came close to, they are not offering to work harder. They are describing a manufacturing process.

Red flag: a guaranteed number of links

The most common promise in the industry and the one that determines everything else about how the work will be done.

Editorial links are decisions made by people who do not work for your provider. Those decisions are unpredictable in timing and in volume. A month with strong outreach and a good asset can produce nothing because a news cycle swallowed the attention; a quiet month can produce six because one journalist happened to be writing something adjacent.

Guaranteeing an exact monthly number eliminates that variance. There is only one way to eliminate it: own or pay for the supply. The guarantee is therefore not a statement about the provider's skill. It is a statement about their inventory.

The damage is not only in the links themselves. A quota creates a hard internal deadline that overrides quality judgment. Late in the month, short of the number, somebody at the agency has to decide, and nobody is going to call and explain that outreach was slow. They will fill the gap, and you will see a report with the right number on it and no way to tell which entries were earned.

Red flag: high-authority links at volume

"Fifty Trust Flow 50-plus links per quarter" or its equivalent in whatever metric the provider prefers. This is where the benchmark does the most work.

Fifty strong referring domains a quarter is two hundred a year. My twenty-year profile earned 446 domains at Trust Flow 41 or above in total — a little over twenty a year, and at a lower threshold than the one being promised. A proposal offering ten times that rate, at a higher quality bar, for a site with no history, is describing something that does not occur naturally in the wild.

What is being described instead is almost always one of three things:

• Sites whose metrics were built deliberately in order to sell placements at a premium. A high score is a manufacturable product; it is computed from the link graph, and the link graph can be fed.
• Sitewide or template links from a small number of domains, which inflate link counts without adding referring domains. Remember that one domain in my profile accounts for 16.8% of all links. Link counts and domain counts are wildly different measurements, and proposals often blur them.
• Metric thresholds measured at the domain level while the actual placement sits on a subdomain, a directory or an orphaned section with none of the parent site's standing.

The question that resolves it: what is the referring domain count, at that threshold, that you have earned for any single client in a twelve-month period, and can I see those domains?

Red flag: .edu and .gov links on demand

Five .edu links in twenty years. Zero .gov. That is what an unbought profile in a competitive commercial niche produced.

University and government sites rarely link to commercial businesses, because they have no reason to. When they do, a real relationship generally exists: a researcher cited something, a library added a useful reference, a local authority listed a supplier or a community sponsor.

Those links are not obtainable on a schedule. Any provider offering them as a menu item is describing one of a small number of things, all of which are worse than they sound:

• Scholarship placements. The scholarship tactic — offering a small award in exchange for a listing on a university's financial aid page — was heavily worked for years and is now widely recognized. Many institutions strip or nofollow these links, and the pages that remain are frequently lists of dozens of unrelated commercial sites, which is a footprint rather than an endorsement.
• Student profile pages, forums and course software. User-generated areas of an .edu domain that anybody can post to. The domain suffix is the only thing of value being sold, and the domain suffix is not itself a ranking factor.
• Abandoned or compromised sections. Old subdomains, unmaintained departmental sites, and pages that a security team will eventually clean.

The .edu and .gov suffixes have a mythology in this industry that outlives all evidence for it. Google has never treated a top-level domain as a quality signal in itself, and a link from a neglected corner of a university site is worth less than a link from a well-read trade publication in your actual sector. When a proposal features these prominently, it is selling the mythology.

Red flag: guaranteed rankings, and guaranteed timelines

Nobody controls the ranking of a page. The provider does not control the algorithm, your competitors' activity, the search demand curve, or whether Google decides to change what the results page looks like for your query. A guarantee of position one, or of a specific percentage traffic increase by a specific date, is a promise about something outside the promiser's control.

In practice these guarantees are honored in one of three ways. First, by choosing the keyword: a guarantee attached to a phrase with no competition and no search volume is trivially achievable and worth nothing, so read the fine print for who selects the terms. Second, by defining success loosely — "top of Google" can mean a map pack, a branded query, a personalized result, or position ten. Third, by refunding, which sounds like the provider carrying the risk and is not: you get your fee back and you keep the link profile, and if that profile is the reason you are in trouble, the refund does not begin to cover the cleanup.

Guaranteed timelines are a milder version of the same problem. "First links within fourteen days" tells you the tactic before you have asked: nothing earned lands that fast.

Red flag: no named prospects, and relationship talk instead of specifics

A proposal should be able to tell you, before any money changes hands, roughly who is going to be contacted. Not the final list — that gets built during the engagement — but a sample. Twenty sites. Real domain names, with a sentence on why each one is a plausible target for your business.

A provider who has done the homework can produce this in an afternoon, and producing it is how they demonstrate they understand your market. A provider who cannot is either planning to work from generic inventory, or has not thought about your business at all.

The substitute you will be offered is relationship language: "we have relationships with top-tier publishers," "our network includes major industry sites," "we work with journalists at national outlets." These sentences are unfalsifiable by design. Test them with one question: name three, and show me the last piece each of them published that included a client link.

Genuine media relationships do exist and they are valuable. People who have them are usually specific and slightly proud about it, because the relationships took years. Vagueness in this particular area is close to diagnostic.

A related tell: the proposal that describes process at length and prospects not at all. Ten pages on methodology, discovery, strategy and reporting, with no indication of which websites will be approached. Process is easy to write. Prospects require research.

What a good proposal looks like

Since this page is largely negative, here is the positive version. A proposal I would take seriously contains most of the following.

• An assessment of the current profile that includes something unflattering. Someone who has looked will have found a problem.
• A named tactic or two, with reasoning tied to your specific situation, rather than a list of everything the agency can do.
• A sample prospect list, with real domains and a line on each.
• A forecast expressed as a range, with the response rate assumption stated, and weighted toward the back half of the engagement.
• Deliverables the agency controls — lists built, assets produced, pitches sent, contacts made — separated clearly from outcomes it does not.
• A maintenance component. On my profile, 717 referring domains at Trust Flow 21 or above have stopped linking entirely, and 97.4% of all lost links died while the source page was still perfectly reachable. Recovery is usually the highest-yield work available, and almost no proposal includes it. One that does has been written by somebody who understands the asset.
• An explicit statement of what they will not do, and what they need from you.

None of that is exotic. It is what a proposal looks like when the person writing it expects to be judged on the work rather than on the document.