Link building proposals are unusually opaque documents. They arrive full of numbers, and almost none of the numbers describe the thing you are actually buying. What follows is a list of the patterns that, in ten years of reading these, have most reliably preceded a disappointing engagement.
A caveat worth stating plainly: several of these appear in proposals from perfectly competent agencies who simply write proposals badly. Treat each one as a question to ask, not a verdict to deliver.
1. A guaranteed domain rating
"All links DR 40+." It sounds like a quality floor. It is a purchasing filter, and one that is trivially easy to satisfy with sites that have a high modelled score and no readers.
Domain rating is a third-party estimate derived from a backlink graph. It can be inflated deliberately, and at the cheaper end of this market it routinely is. A DR 55 domain with 300 monthly visitors exists in large numbers precisely because people buy on DR.
Ask instead: what is the minimum verified organic traffic on a referring domain? A real answer is a number. Ours is 1,000 sessions a month.
2. A monthly link count with no gap analysis behind it
"Twelve links per month." Twelve compared to what? The only defensible way to arrive at a volume is to pool the referring domains of the URLs currently outranking your page, subtract yours, and size the programme against the difference.
A proposal that names a number before running that subtraction has derived the number from its own capacity, not from your situation.
Ask instead: what is my referring-domain gap on the target page, and how did you calculate it?
3. A sample list you are not allowed to verify
Some agencies will show you a list of publications they "have relationships with" but decline to show live placements, citing client confidentiality. Confidentiality is real and NDAs are normal. But live URLs are public documents — anyone can read them — and an agency that cannot produce twenty from the last quarter is telling you something.
Ask instead: twenty live URLs from placements made in the last ninety days. Client names can be withheld; the URLs cannot be.
4. Pricing well below the market floor
The 2026 market average for a quality editorial link is around $500. Credible SaaS placements run $150 to $500, premium publications $700 to $1,500 and above. Below roughly $150, the arithmetic simply does not close: nobody can prospect, score, pitch, write a publishable draft and compensate an editor at that price.
So the inventory is coming from somewhere else — typically a network of sites built to sell links, sharing hosts, templates and outbound patterns.
| Quote per link | What it usually means |
|---|---|
| Under $100 | Network inventory, near-certainly |
| $100–$150 | Marketplace resale, low or no vetting |
| $150–$500 | The credible band for SaaS placements |
| $700–$1,500+ | Premium publications, digital PR outcomes |
5. No mention of your category's comparison surface
For a SaaS company, some of the highest-value placements are not blog posts at all. They are round-ups, "best tools for X" lists, alternatives pages, integration directories and review platforms — the pages a buyer opens two weeks before shortlisting.
A proposal that never mentions them was written from a generalist template. It may still produce decent links. It will not touch the surface where your buying decisions actually get influenced.
6. Reporting described as a "monthly report"
Ask what is in it. If the answer is a summary document delivered at month end, you are being offered a narrative rather than a record. The alternative — a live sheet with one row per placement, listing URL, anchor, surrounding paragraph, do-follow state, referring-page traffic and index date — lets you spot-check the work on any day you like.
The difference is not presentational. One can be audited; the other has to be believed.
7. No replacement warranty
Links disappear. Editors revise, sites get sold, pages get pruned, a do-follow quietly becomes a no-follow. Industry-wide, a meaningful share of placements are gone within two years.
A proposal with no clause covering this has moved the entire risk onto you. The clause you want specifies a monitoring window (twelve months is standard), what triggers a replacement (removal, no-follow flip, de-indexation, 404) and that replacement is free.
Warning phrase: "that doesn't really happen in practice."
8. Vagueness about who writes the content
If your name is going on an article, you should know who drafted it. "Our content team" can mean three staff editors with software marketing backgrounds, or a marketplace with a two-day turnaround and no subject knowledge.
Ask instead: can I meet the writer assigned to my account, and can I see two pieces they have written in my category?
9. Agreement with everything in your brief
The subtlest one. A proposal that endorses every assumption you made has not examined them. Somewhere in a good response there is a paragraph that says, politely, that point four is wrong.
We once lost a pitch to an agency whose opening line was that the client's target keyword had almost no commercial search volume and they should choose a different one. They were right and we had missed it. That is what a good proposal reads like.
The three-question version
If you only have ten minutes, these do most of the work.
Where does the inventory come from? A method, not a price list.
What did you reject last month, and why? Almost nobody keeps a rejection log. The ones who do are usually worth talking to.
What happens if a link is gone in month eight? A written warranty, or an excuse.
What is not a red flag
A few things get treated as warning signs and should not be. A three-month minimum term is reasonable — outreach sent in month one publishes in month two and is measurable in month four, so anything shorter tests the sales process rather than the work. Refusing to guarantee rankings is correct, not evasive; nobody controls a third-party ranking system. And declining your enquiry outright is a good sign, not a bad one: an agency that takes every client has no criteria.