How bklink values a backlink and a property

bklink estimates what a backlink would sell for today, then discounts it for everything an audit can see that makes it worth less: how the link is tagged, whether the page still exists, how many other links the same site gave you, whether the same article was republished elsewhere, and whether the site shares a server with a network of others.

This page describes the model exactly as the audit engine runs it. Every rate, multiplier and threshold below is read from the engine's own code when this page loads, and the worked example is computed by the engine itself. A valuation is an estimate of what comparable links sell for, not an appraisal — not a prediction of rankings, traffic, or what anyone will pay.

The two numbers in every report

Gross value

What the links would cost at market rates, before bklink discounts anything for hosting or syndication. It is shown with a low–high range, because market rates vary.

Estimated value

What the profile is realistically worth after every discount below. Links the audit could not verify are left out of it entirely. This is the figure on the dashboard, on project cards and in the value-over-time chart.

What the model reads

• The link itself — bklink loads every page you list and reads the link: followed, nofollow / sponsored / ugc, or missing. Nothing here comes from a third-party index.
• Domain Rating — Ahrefs Domain Rating, from the Ahrefs API, and nothing else. A domain Ahrefs does not rate is unrated, not estimated. Cached per domain for up to 30 days.
• Organic traffic — DataForSEO's estimate of the linking site's monthly organic visits. Cached per domain for up to 30 days.
• Content category — bklink's own classification of the site from its page metadata (title, description, keywords). Cached for up to 7 days.
• Hosting — every linking domain is resolved to the address it serves from, and that address to its hosting network, via public DNS and routing data. Cached for up to 7 days.
• Indexability — whether the page tells search engines not to index it: a noindex tag or header, a canonical pointing elsewhere, or a robots.txt block.

Step 1 — Is there a link worth anything?

Some links are worth $0 before any rate is applied, and the report says why for each one:

• No link — the page loads, but the link you paid for is not on it.
• Dead — the page returns 404 or 410, its domain no longer resolves, it serves a "not found" page with a 200 status, or it returned a server error on two audits in a row. A single server error, a block (401/403) or a timeout is not treated as dead.
• Not indexable — the page asks search engines not to index it, so the link passes nothing.
• Search-engine domains — links from google.com, bing.com, yahoo.com and their news hosts are crawl artefacts, not placements.
• Unrated — a domain with no Ahrefs Domain Rating has no rate to look up.

Unverified links — pages that blocked or timed out our auditor — are handled separately (step 4).

Step 2 — The market rate for the link's authority

Each link gets a low, mid and high rate from a table keyed on Domain Rating. Followed and nofollow links use different tables: a nofollow link passes no direct ranking credit and is bought for referral traffic and brand exposure instead, so it prices off a much lower base. The mid rate is the one carried forward.

Step 3 — Adjusted for the site's audience and niche

Domain Rating alone is a weak predictor of what a link sells for — across marketplace data it correlates below 0.5 with price. Two adjustments close the best-evidenced gaps. Both multiply the rate; missing data multiplies by 1, so a link is never marked down for something we have not measured.

Categories without repeated, quantified price evidence — news, sports, gaming, entertainment, travel, food, automotive, real estate, fashion, home and garden — are deliberately left at 1× rather than given an invented number.

Step 4 — Links the audit could not verify

When a site blocks or times out our auditor, we cannot see whether the link exists or how it is tagged. Such a link is priced as a band from $0 (there may be no link at all) up to the followed-link high rate (it may be a full followed link). Its midpoint uses the mix this same audit actually verified — if 13% of the links we could read were followed, the midpoint assumes a 13% chance this one is too. Unverified links count toward gross value, and are disclosed separately, but are left out of the estimated value entirely.

Step 5 — Several links from one site

Ten links from one site are not ten sites vouching for you. Within each linking domain, links are ranked by what they are worth:

A link already worth $0 — dead, missing or not indexable — takes no place in that ranking, so it never pushes a live link down.

Step 6 — The same article on many sites

A press release distributed over a wire appears on hundreds of sites as one article, and search engines consolidate it rather than counting every copy. bklink groups copies of the same article — the same canonical address declared by the page, or the same article path across hosts — and keeps the strongest copy at 100%. Every other copy keeps 5%: a wire pickup is a real page that can send visitors, but it adds almost nothing in search over the original. This applies across different domains; step 5 covers repeats within one.

Step 7 — Hosting

Each linking site is resolved to the server address it runs on. When 10 or more distinct linking domains share one address, bklink treats them as one operator's network — a private blog network — and each of those links keeps 5% of its value. Adjacent addresses in one announced network block are treated as one server when the domains are concentrated on them (8 or more domains per address), because one operator often sits on several neighbouring addresses.

Content delivery networks (Cloudflare, Akamai, Fastly) and the large clouds are never grouped this way: their addresses are shared by thousands of unrelated sites by design, and calling that a network would be a false accusation against the publisher.

Every other link keeps 50%. That baseline is deliberate: the rate tables describe a standalone editorial link at list price, and the typical paid placement is worth materially less than list.

Worked example

Three illustrative links, run through the model exactly as an audit runs them — a followed link and a nofollow link on the same technology site, and a wire republication of the same article on a small news site.

Market factor = traffic multiplier × category multiplier. The nofollow link is the second link on its domain, so it keeps 20% of its rate. The wire copy is the same article as the first link, so it keeps 5%. Neither site shares a server with a network, so both keep 50% for hosting.

How a property is valued

A property — the site you are building links to — is not the sum of its reports. bklink takes the latest audit of every campaign on the property, merges their links, removes duplicates (the same address listed in two campaigns counts once), and runs the whole model again on the combined set.

That matters because steps 5 to 7 depend on everything else in the set. A link that was the strongest from its domain in one campaign can become the second once another campaign's stronger link from the same site is merged in, and a server that looked independent in one audit can cross the network threshold once all of a property's links are counted together. The property figure is the estimated value of that combined set.

Value over time replays that calculation at each audit: at every point on the chart, the value is what the property's links would have been worth using the latest audit of each campaign as of that moment — not an interpolation between them.

What the model does not capture

• Placement on the page. An in-content link and a footer or sidebar link are priced the same; the auditor reads the link's tags, not where it sits.
• Directional estimates. The nofollow table and the same-domain curve come from published market research on link pricing, not from bklink's own transaction data.
• Vendor models. Organic traffic is DataForSEO's estimate, not measured visits; Domain Rating is Ahrefs' model of link authority.
• Category is approximate. It is read from each site's own metadata and can misfile a site whose keywords describe one section rather than the whole.
• Language and market scarcity — the premium some non-English markets command — is not modelled; the evidence for it was too weak to isolate.

When the model changes

Reports are snapshots. A report you have already shared keeps the figures it was built with; a new audit uses the model as it stands. The most recent changes:

• 20 September 2026 — only Ahrefs Domain Rating prices a link. It replaced a vendor rank that ran about 19% low on the same links.
• 14 September 2026 — the network threshold rose to 10 domains on one address, and network links keep 5% in the estimate rather than nothing.

Questions about a figure in your report: support@bklink.uk. Written and maintained by Palash Bagchi, founder of bklink. See also About bklink.