Discussion
  • Agency

Are link-building packages sold by link volume misaligned with client interests? How would you structure them instead?

Published 1 reply

Discussion starter: an agency question, posted with an answer from Niraj Raut to open the thread. If you have dealt with this on a site, reply with what you saw, especially where it differs.

Disclosure: Niraj Raut, who posted this answer, runs the SEO consultancy linked at the end of it.

1 reply

  • admin
    Admin
    Original poster

    Mostly, yes. A package priced per link pays the agency for the unit that is cheapest to produce and easiest for Google to discount, while the client carries the policy risk and the opportunity cost. The qualification is that counting is not the real problem; an undefined unit is. A count of placements that each meet written, auditable criteria can be a fair delivery measure. So I would change the unit and the risk split rather than abandon numbers: a base fee for strategy and campaign work, a placement fee paid only for links that meet a published definition, and a quarterly review that compares pages that received links with comparable pages that did not.

    Where the misalignment comes from

    Once the unit is “a link above a metric threshold”, three incentives pull apart.

    • Margin rewards the cheapest link that qualifies. If the price is fixed per link above, say, a DR 30 floor, the agency earns more by finding the cheapest sites that clear the floor. Sites that openly sell posts clear it cheaply. An editorial link from a publication the client’s buyers read costs far more effort. The package pushes towards links that are unlikely to be worth anything.
    • The client holds the risk. Google’s spam policies list buying or selling links for ranking purposes as link spam, including exchanging money for links or for posts that contain links, along with links with optimised anchor text in articles, guest posts or press releases distributed on other sites. If those links are discounted or lead to a manual action, the consequences land on the client’s domain. The agency’s exposure is a cancelled contract.
    • The easy targets are not the valuable ones. The pages a business needs to rank (categories, services, product lines) are harder to earn links to than blog content. A package that doesn’t name target pages tends to route links to the homepage and blog posts, because that is where they are easiest to place.

    The metric threshold makes it worse. Domain Rating and Domain Authority are vendor models of the link graph; neither is a Google metric, and a site that sells posts can hold a healthy score. A floor filters out obvious junk, but it does not identify the links Google is likely to count.

    Comments from Google point the same way. Search Engine Roundtable reported, from an attendee’s account, that Gary Illyes told SERP Conf in April 2024 that Google needs “very few links to rank pages”, and that he later posted that he shouldn’t have said it. I would read that as a direction rather than a specification, but it is hard to square with paying for volume.

    Hypothetical example:

    A retailer buys 15 links a month with a DR 30 floor. Six months later the report lists 90 links: 70 on sites with public “write for us” price lists, 12 on pages that don’t appear in Google when you search a quoted sentence from them, and 8 from trade publications that covered the retailer’s pricing survey. The 8 point at the survey page. None point at the category pages that earn revenue, and the survey page doesn’t link to them either. Under a qualifying-placement model the agency would have been paid for the 8, and the brief would have required the survey page to link to the categories it supports.

    Why paying for rankings does not fix it

    The obvious alternative is to pay for outcomes: positions, traffic or revenue. That swaps one misalignment for three others.

    • Attribution. Rankings move with core updates, competitor changes and the client’s own releases. The agency gets paid for, or blamed for, movement it didn’t cause.
    • Target selection. When fees attach to positions, the easiest terms to move become the most attractive, whatever their commercial value.
    • Speed over safety. A monthly performance fee rewards whatever moves rankings soonest, and that pressure makes paid links with optimised anchors tempting, which is the risk the client was trying to avoid.

    Outcome measures belong in the review rather than the unit price, unless the contract includes a control group and a clawback.

    Pricing models compared

    Model Client pays for Incentive it creates Where it fits
    Per link above a metric floor A count of links above a DR or DA threshold Find the cheapest sites that clear the floor Rarely; only with a strict definition and audit rights
    Per qualifying placement Links that meet written criteria, priced by publication tier Fewer, better links; negotiation shifts to the criteria Clients with clear target pages and something worth covering
    Campaign or retainer Research, assets and outreach cycles Deliver activity; the client carries the output risk Digital PR where hit rates are uncertain and coverage has value beyond links
    Hybrid: base fee plus qualifying placements Strategy and campaign work, plus a fee per qualifying placement Both sides share the output risk Ongoing programs that mix campaigns with steady outreach
    Outcome fee Movement in rankings, traffic or revenue Chase easy terms and fast tactics; disputes over attribution Only with a control group, a long window and a clawback for policy breaches

    Define a qualifying placement before you price one

    The definition does more alignment work than the pricing model. It turns “a link” into something the client can audit and the agency cannot cheaply fake. A workable one:

    • The publisher chose to link, and no money, product or service went to the publisher in exchange for a followed link. Paid or gifted placements can still be worth buying for exposure, but Google’s policy says links bought for advertising or sponsorship are not a violation as long as they carry rel="sponsored" or rel="nofollow", so they are qualified, reported and priced separately.
    • The linking page is relevant to the target page’s topic or to the client’s buyers.
    • The site publishes for readers: no public price list for posts, no pattern of near-identical sponsored posts for unrelated brands, and an audience trend that is not a spike followed by a collapse.
    • The page is indexed: searching a quoted sentence from it returns the page.
    • The link points at a target URL agreed in advance, with anchor text the publisher wrote.
    • It is still live and unchanged after 90 days, or the agency replaces it at no cost.

    Then price by tier, for example national or major trade media, specialist publications with an engaged audience, and relevant smaller sites and partners. The tier sets the price; the definition decides eligibility.

    The structure I would use

    Base fee

    Target-page selection, a link gap analysis against the pages that outrank you, asset and story development, and outreach. Fixed monthly, because this work happens whether or not a campaign lands.

    Placement fee

    Paid per qualifying placement and priced by tier, with a quarterly minimum rather than a monthly one and a make-good if it is missed. A quarterly minimum absorbs the lumpiness of campaign-led links.

    Quarterly review

    Linked pages against holdout pages, referral sessions and conversions from placements, coverage and brand mentions, and a decision on where the next quarter’s links should point.

    Contract terms

    • A placement log: URL, date, target, anchor, rel attribute, and anything paid or gifted
    • Client sign-off on target pages and on any paid placement
    • No undisclosed paid followed links, private networks or link exchanges
    • If placements the agency built lead to a manual action for unnatural links, the agency funds the clean-up
    • The client owns the assets, data and publisher relationships its budget paid for

    This keeps what clients like about volume packages, a predictable number to hold the agency to, while making each unit worth counting. It also gives the agency a fair deal: the base fee covers work that doesn’t always produce links, and the placement fee rewards the links that matter.

    How to tell after two quarters whether the links did anything

    A review should compare, not just report. Before the program starts, take a set of comparable commercial pages on the same template with similar traffic and positions, and split them at random: one group receives links, the other receives none for the period. Compare non-brand clicks and visibility for each group’s queries.

    Two quarters rather than one, because links have to be discovered and the linking pages crawled before they can count, and because a single quarter can overlap a core update: Google completed core updates in both March 2026 and May 2026. A holdout group absorbs update effects that hit both groups equally, and keeping both groups on one template stops a template-specific update effect from passing for a link effect.

    If linked pages do not pull ahead of the holdout after two quarters, links probably were not the constraint on those pages, and the budget belongs in content, internal linking or the offer. If they do, you have an effect size to price the next year against.

    The practical position

    If I were buying, I would decline any package whose unit is “a link above DR X”, and accept a count only when each unit meets a written definition I can audit. If I were selling, I would price the work and the qualifying outcome separately, because that is how an agency gets paid fairly when a strong campaign produces fewer, better links than a weak one.

    Need help with this on your own site?

    Niraj Raut works with businesses in Nepal, Australia, the UK, Europe and the US on technical SEO, ecommerce SEO, local SEO and AI search.

    See SEO services

Add a reply

Replies are open to approved members. Log in or apply to join.