Price is a weak signal in link building. It is not a meaningless one — below roughly $150 a placement the arithmetic stops working and the inventory is coming from somewhere you will not like — but above that floor, cost tells you remarkably little. Two agencies quoting $400 and $900 per link are often buying from the same marketplace and differ mainly in overhead.
What actually varies is harder to see from a proposal, which is why it needs to be asked about directly. Six things.
1. Whether the target list is derived or invented
A good agency starts by subtracting. It pools the referring domains of the URLs currently outranking your page, removes yours, and treats the remainder as the scope of work. The number that falls out — often forty to a hundred and twenty domains in a mid-competition SaaS category — determines the volume, the timeline and the budget.
A weaker agency starts from its inventory. It has relationships or purchase agreements with a set of sites, and the proposal is an arrangement of those sites into a monthly number. The links will be real. They will simply have very little to do with why your competitors outrank you.
How to check: ask for the gap analysis before signing. Some will run a sample for free. If the answer is that gap analysis happens after onboarding, you are being sold inventory.
2. Whether anyone reads the publications
Domain rating is a modelled score. It can be inflated, and in the cheaper end of this market it routinely is. What cannot be faked easily is a readership: verifiable organic traffic, a topic history predating your outreach, a named author with a professional footprint, an editor who replies.
The SaaS-specific version of this matters more than the general one. A DR 70 general-business blog with 40,000 monthly visits, none of whom evaluate software, is worth less to you than a DR 45 operations publication read by exactly your buyer.
How to check: ask for twenty live URLs from the last quarter and run each host domain through a traffic tool. Then read three of the articles. You will know within ten minutes.
3. Whether the agency can say no
Every agency has a rejection rate. Almost none publish it. Ours runs around seven in ten prospected domains, and we show clients the log with reason codes each month — partly because it is the clearest evidence of what the retainer buys, and partly because it stops us quietly lowering the bar when a month is running behind.
The reason this matters commercially: an agency with a monthly link quota and no rejection discipline will hit the quota. It has to. The only variable left is quality.
How to check: "Can I see what you rejected last month, and why?" It is a slightly rude question and the response tells you a great deal.
4. Whether it understands the comparison surface
This is the difference between a general link building agency and a SaaS one. Your buyer does not search your brand — they search the category, then "best tools for X", then "alternatives to" whoever they trialled first. Those pages belong to publishers, and appearing in them is a placement workstream with its own process.
A generalist will not mention round-ups, alternatives pages, integration directories or review platforms in the proposal, because they are not part of the standard playbook. A SaaS specialist will lead with them, because they produce referral demo requests immediately, independently of any ranking effect.
How to check: ask how many round-ups in your category they have audited, and how many of your competitors' listings they can name. Vagueness here is disqualifying.
5. Whether the reporting can be checked
A monthly PDF summarising activity is a document about the agency. A live sheet listing every placement with its live URL, anchor text, surrounding paragraph, do-follow status, referring-page traffic and first index date is a document about the work.
The distinction is not stylistic. With the second, you can spot-check five links on any given Tuesday. With the first, you are approving a narrative.
| Column | Why it matters |
|---|---|
| Live URL | Lets you verify existence in one click |
| Anchor text | The only way to track profile drift over twelve months |
| Surrounding paragraph | Distinguishes a contextual link from a bio-line link |
| Do-follow status | Some placements flip after publication |
| Referring-page traffic | Distinguishes a read page from an indexed one |
| First index date | Unindexed at day 30 means it does nothing |
6. Whether it will tell you not to buy
The most useful thing an agency ever said to us, back when we were on the client side, was that our category page would not convert the traffic even if it ranked, and we should fix the page before spending anything on links to it. They talked themselves out of two quarters of revenue and earned four years of it instead.
An agency that agrees with every premise in your brief has not examined your brief. Somewhere in a good proposal there is a paragraph that begins with a polite disagreement.
What price does tell you
Not everything, but not nothing. The 2026 market average for a quality editorial link is around $500, up roughly 45% from $350 in 2022. Credible SaaS placements generally run $150–$500; premium publications $700–$1,500 and above. Guest posts average $220–$609, link insertions around $141, and paid inclusion in a category round-up $350–$750.
A quote materially below those bands is not a bargain — it is a different product. Ask where the inventory comes from and how many other sites share its footprint.
The uncomfortable summary
Most of what distinguishes a good agency is process discipline that costs money and produces no visible artefact. Rejecting seven domains in ten is expensive. Verifying traffic on every candidate is expensive. Writing a draft properly rather than briefing a content mill is expensive. None of it shows up in a link count.
Which is why the link count is the wrong thing to compare proposals on, and why the agencies that compete on it are the ones you should read most carefully.
Compare on method, verify with live URLs, and ask the awkward questions early. The market is not short of competent operators; it is short of clients who ask them anything difficult.