Every supplier in this market says roughly the same things. High-quality links. Manual outreach. White hat. Real websites. Since they deliver measurably different products, the language is not doing any work — which means you have to identify the model some other way.
The four models
| Model | You are buying | Good at | Breaks when |
|---|---|---|---|
| Marketplace | Inventory from a catalogue | Speed, price transparency, one-offs | You need vetting or strategy |
| Productised service | Fixed output to a published spec | Predictability, auditability, budgeting | You need strategy owned end to end |
| Agency retainer | Bespoke programme with thinking attached | Complex situations, senior input | Scope drifts and cannot be reconstructed |
| Freelancer | One person's relationships | Deep niche access, low overhead | Capacity, redundancy, infrastructure |
None of these is wrong. A marketplace is honest about being a catalogue, and if you have the in-house expertise to vet, it is efficient. The problem is buying a marketplace while believing you bought a programme.
Six questions that identify the model
1. "How did you arrive at the monthly link count in this proposal?"
A programme derives it from a gap analysis — the referring domains of the pages outranking you, minus yours. A marketplace derives it from your budget divided by an average price.
Listen for whether the answer references your competitors' numbers. If the volume was chosen before anyone looked at your SERP, you are buying capacity rather than a plan.
2. "What did you reject last month, and why?"
The most revealing question available. A supplier doing genuine prospecting rejects roughly two thirds of candidates and can tell you the reason codes. A supplier drawing from a fixed catalogue has nothing to reject — the vetting happened once, years ago.
3. "What is your minimum verified organic traffic on a referring domain?"
You want a number and a named third-party index. A domain-rating threshold instead tells you the filter — and a DR filter selects for sites built to satisfy it.
4. "Show me twenty live URLs from the last quarter."
Published articles are public documents. Client names can be redacted; the URLs cannot reasonably be withheld. Then check three of them yourself: does the page have traffic, is there a named author, is the link contextual.
5. "When does a placement become billable?"
The answer you want: once live, indexed and verified. Billing on outreach sent or placements "secured" moves the entire execution risk to you, and it is a common structure precisely because most buyers do not ask.
6. "What do I keep if I leave?"
Gap analysis, target list, content you paid for. If those stay with the supplier, switching means paying someone else to rebuild six to nine months of work you already funded.
The tell that shows up in the first reply
Send question 1 by email and read the response for one thing: does it contain any of your numbers?
A supplier who has looked at your SERP will mention your competitors' referring domain counts, or your current position, or the fact that your category is dominated by round-ups. A supplier selling inventory will send a package tier.
It takes one email and it sorts most shortlists.
What to specify before you talk to anyone
Write this down first. It takes ninety minutes and it changes the conversation from "what do you sell" to "can you do this".
- The exact URL you need to move, and the term it should own.
- Who is above you and their referring domain counts. One line: "we have 19, they average 78."
- What has been tried, including anything you would rather not admit — it determines whether quarter one is acquisition or cleanup.
- Your budget, stated. Withholding it produces a proposal reverse-engineered from what they think you will pay.
- Who approves drafts and how fast. This is the single biggest determinant of programme speed.
- What you will not accept — publications, competitor adjacency, whether paid inclusions are acceptable.
Structuring the engagement
Four terms worth insisting on regardless of model.
Three-month minimum, not twelve. Three is reasonable — outreach sent in month one publishes in month two and is measurable in month four. Twelve with no exit is a term to decline.
Billable only when live, indexed and verified. The single clause that reallocates most of the risk.
Twelve-month replacement warranty with defined triggers: removal, no-follow flip, de-index, 404. Replacement, not a credit note.
Ownership transfers on payment. Gap analysis, target list, content.
Managing it once it runs
Three habits, none of which take more than twenty minutes a month.
- Spot-check five random placements monthly. Random matters — let the supplier see you picking. Open the page, find the link, check the rel attribute, confirm it is indexed.
- Ask which placement was weakest. There always is one and they know which. A supplier who says every placement was equally strong has stopped assessing their own work.
- Track the rejection rate against delivery. If it dips in months where volume was behind, the quota is setting the standard rather than the other way round.
Clients who spot-check are the ones suppliers work hardest for. That is not cynicism about the industry — it is how every service relationship works, and it costs you twenty minutes.
When not to outsource at all
Three situations where the honest answer is to keep the money.
You have not claimed what is already available. Integration partners, associations, conferences, customers with case study pages. Most companies have twenty to fifty unclaimed links sitting there. Two weeks of email beats two months of retainer.
Your budget is under about $1,500. Below that, volume outreach cannot produce enough to matter. Build one citable asset instead, or spend the quarter on internal linking.
The page will not convert the traffic. Ranking a page that converts at 0.3% is an expensive way to learn the conversion rate.
A supplier who tells you one of these things on the first call is worth more attention than one who does not, even though — especially though — it costs them the sale.