Link Audit: Donor Metrics, Anchors, and Budget

A link audit is one of the most frequently skipped procedures: budgets are spent by guesswork, and results are explained away by updates. Below is a practical methodology: what an audit includes, which donor metrics actually mean something, how to calculate the anchor cloud and budget, and how working with link networks works in practice. With case studies and figures.

When to Run an Audit and What It Delivers

The ideal time is before building a website. Assessing competitors’ link profiles shows whether it’s worth entering a niche at all: you could invest in development, content, and technical optimization, then discover that competing in that GEO requires roughly a million rubles a month just for links. These situations aren’t hypothetical—they come up regularly, usually after the budget has already been spent on everything else.

More often, an audit is done on a project that already has some traction. In that case, analyzing competitors is accompanied by a review of the site’s own profile, and the next step is to make adjustments.

What an audit covers:

• the volume and quality of links needed to reach target positions;
• an approximate budget, including a month-by-month breakdown;
• a list of pages and anchors where link buying will have the greatest impact.

One limitation to keep in mind from the outset: services show current prices, and there’s no way to reconstruct competitors’ historical spending. A site that bought links several years ago at the prices back then benefits today from results that would cost you considerably more.

Competitor Selection: Where the Whole Calculation Can Go Wrong

If competitors are chosen incorrectly—different scale, different GEO, different type of site—all the subsequent figures will be wrong, and the audit will lead you astray.

Selection criteria:

• the closest possible match in topic; for an online store, also match the product range and categories;
• Google traffic specifically in the GEO you’re targeting;
• the sample should include at least one site with more traffic than yours—as a benchmark for where to aim—and several at roughly your level, to understand what link profile corresponds to your current traffic volume.

A working sample consists of three or four sites that then undergo a full analysis of link volume and quality, and estimated costs.

Where to Get Data on Your Own Donors

Compile the donor list by combining two exports—from Ahrefs and Google Search Console. Each source has its drawbacks, and together they fill in each other’s gaps.

Ahrefs remains the best tool for working with links, but more and more sites are hiding their data from it, and there are more such sites every year. Search Console shows links regardless of what webmasters want—you can only hide from it by dropping out of Google’s index. But it has two drawbacks: the report is limited to roughly a thousand domains, and it doesn’t distinguish between dofollow and nofollow. You have to check the link type manually by visiting the final URL, sometimes through two or three intermediate pages. It’s tedious work, but the resulting picture is more accurate.

There’s also no way to automatically export specific URLs from Search Console: you can export by domain, but getting down to individual pages requires manual work. Data Studio reports don’t solve the problem. The overall trend is that there’s less data in both analytics and Search Console as the years go by.

What to use: if links are only an occasional task, Semrush is a cheaper option. If this is your main line of work, there’s still no alternative to Ahrefs—for a shared subscription, it costs around a hundred dollars, which seems pretty reasonable now that limits have been introduced. Majestic and Serpstat still lag behind in how up-to-date their link indexes are.

Donor Metrics: What to Check Besides DR

A telling example from a real spreadsheet. A domain with DR 73 looks like a great site at first glance. Then you look closer: it has only 49 incoming domains, and just 24 of them link via dofollow. And the number of outgoing domains in its links is 401.

In theory, a domain could get DR 73 with fifty or so incoming domains if giants like Amazon or Google were linking to it, but checking the other metrics doesn’t bear that out. This donor falls into the “bad and weak” category: it has nothing to pass on, because it has orders of magnitude more outgoing than incoming domains.

The ratio of incoming to outgoing domains is the main filter. Not absolute figures, but the proportion itself. The more quality sites link to a donor and the fewer sites it links out to, the stronger it is.

A practical set of metrics for the spreadsheet: DR, total incoming domains and dofollow incoming domains separately, outgoing domains in links, IP, traffic, CF and TF from Majestic, spam metrics, and Checktrust data. For competitors, also count total incoming links, the number of dofollow donors, and, separately, the number of donors linking via a dofollow link within the content—these are the most valuable.

There’s a nuance about nofollow that people often only partly understand. The recipient doesn’t receive link equity through such a link, but some equity still leaves the donor page: it isn’t redistributed among the remaining links, it simply disappears. Google closed this loophole back in 2009, when it announced that using nofollow attributes doesn’t “sculpt” the equity within a page. That’s why outgoing nofollow links are still taken into account when evaluating a site.

What You Can’t Replicate

You can’t copy a competitor’s entire profile: some of their links can’t be bought. An online store might ask a supplier or dealer for a link, or someone might link to them voluntarily. In the Russian-speaking market, though, the paid portion is fairly easy to replicate—almost everything goes through link marketplaces, and services can identify whether a site is listed on one.

Redirect chains should also be checked: domains that have been merged into another site and chains of redirects. They provide a noticeable boost and still work if configured correctly, but often aren’t visible in Ahrefs. One indirect sign is a sharp increase that doesn’t coincide with update dates. Links are hidden en masse in competitive niches, so this step shouldn’t be skipped when analyzing major topics.

The Anchor Cloud Instead of Counting Repeats

A common evaluation method is to count repetitions: five uses of one anchor is fine, ten is already over-optimization. This method is flawed because it ignores the scale of the profile.

It’s better to calculate the share: what proportion of the overall anchor cloud each anchor accounts for. The team’s guideline is no more than five percent for a single anchor, but this isn’t set in stone—there have been cases where ten percent caused no problems. You need to compare against competitors in your niche.

Both extremes are bad. An imbalance where links use one or two anchors and all the others are missing looks unnatural. But a profile where every anchor appears exactly once is also suboptimal: the anchors are simply underused, even though they’re how a search engine understands what a link is about and where it leads—this is stated directly in Google’s documentation on links.

This leads to a position that differs from popular recommendations: anchor links should predominate, and a share above ninety percent is fine. Naturalness comes not from the proportion of anchorless links, but from the variety of the anchors themselves: higher-frequency ones are used more often, lower-frequency ones less often. If over-optimization does occur, it can be diluted with crowd links using plain URLs and brand anchors.

It’s important to note that percentages are controllable. With each month of link buying, the shares in the cloud change: some anchors are emphasized, others aren’t used at all, and the imbalance gradually evens out.

How the Budget Is Calculated

No method gives an exact answer, but you can get close.

The main method. Competitors’ donor lists are checked using Link Detective Pro or seorestore.com—these services maintain downloaded marketplace databases and show where a specific site is sold and at what price. Then the average cost and total number are calculated. Database accuracy ranges from forty to eighty percent: some sites don’t sell links at all, even though a link from them still has value and works.

A rough estimate. Multiply the number of competitors’ dofollow donors by the average link price for the niche and GEO. It’s a blunt method, but good enough for an initial discussion of the ballpark figures.

If the calculated budget is immediately unaffordable for the client, the total is divided into a comfortable monthly amount to arrive at a timeline. That’s more honest than promising results on a budget that isn’t enough for the niche.

Where to Save

• The same site is often listed on several marketplaces at different prices. The owner keeps an eye on the one with better sales and raises the price there, while forgetting about the other one. Lookup services show the lowest price—you should buy there.
• Discounts for volume and loyalty. If you regularly work in one GEO and keep your own list of sites, it makes sense to ask for volume-based terms or a loyalty-program discount.
• Undervalued sites. Topical traffic matters, but it’s overvalued: a site with good link metrics and modest traffic is often cheap and performs better than a well-promoted brand with nothing but high visitor numbers.

Selecting Pages and Anchors for Link Buying

Once both profiles have been analyzed, it becomes clear exactly where to invest. Selection criteria:

• page traffic—current or estimated based on the search volume of the main keyword;
• keyword difficulty;
• SERP analysis: how many links the sites ranking at the top for this query have, and what their metrics are.

You should work on the pages closest to the top—they’re cheaper to push up. A query languishing beyond position one hundred means either extreme competition or problems with textual or technical factors, and links won’t fix it.

What Should Be Left After an Audit

A report you can use, not a presentation. It should include:

• a written section: what was found, what to focus on, and what to stop working on;
• an anchor list with percentage shares;
• a table of your site’s donors with verified metrics, split into strong and weak;
• a comparative table of competitors;
• link exports—your own and competitors’—so the client can verify the conclusions;
• examples of bad, weak, and good links, with an explanation of why they fall into each category;
• a budget estimate and a growth plan, with sources and link-buying costs.

Examples make the report easier to read. A bad link might be some analytics service that latched onto the site by itself; this kind of link profile is compared to barnacles on a ship’s hull that need to be scraped off periodically. A weak link is a typical crowd link: it has a slightly positive effect. A good link comes from a site that has passed every check.

What It Costs: Case Studies with Figures

Actual figures are more useful than any speculation about “how many links are needed.”

• SaaS, US. One hundred links from a network and fifteen through outreach. Around seven thousand dollars for network placements and ten thousand for outreach. The result was reaching position four in a competitive niche and GEO.
• The most successful project. 114 links from a network, 489 from marketplaces, and 17 through outreach, with a total link budget of around twenty-seven thousand dollars spread over several years. The result was around 1,2 million visits. The trend is telling: while they worked only with marketplaces on a budget of two to three hundred dollars, traffic grew slowly and stalled at a couple thousand visits. After network links were added, growth became noticeable after about a month and continued to accelerate. The placement rate was five to ten links a month; outreach was added right at the end.
• Automotive company, CIS. 26 network links and 34 marketplace links, with a budget of 150 thousand rubles over several months—enough to reach top positions.
• SaaS, US, another project. 52 network links, around 3800 dollars. The result was more modest than the previous ones, but positive.

What all the cases have in common: the budget is spread across months and years, not paid all at once.

Working with Link Networks

The criteria for evaluating sites in a network are the same as for ordinary donors, plus checking dropped domains for spam history and signs of filters. Then come the things specific to link networks.

The Main Quality Criterion

The number of unique domains linking to the site via dofollow—and specifically with a “good link.” Traffic is a nice bonus, but it isn’t the deciding factor.

What does not count as a good link, even if the domain is respectable: a page with five hundred outbound links; hacked comments; giant roundups like “one hundred places to visit,” where your link is number one hundred and one.

How many links to place from one site

The practical limit is no more than ten per project, and the fewer, the better. The reason is that several links from one domain do not increase the total weight: the value is divided among the directions. The analogy used by the author of the methodology is a barrel with a tap: no matter how many holes you drill, there will not be more contents.

The logic here is simple. If weight multiplied with the number of links, it would be possible to manipulate search results from a single site. This probably did work in the early days of search engines, but the loophole was closed long ago.

Outbound link control

They monitor not the absolute number but the ratio: the number of unique outbound domains should not exceed the number of inbound domains, with a margin of around ten percent. Once the threshold is reached, placements on the site are stopped—they wait until a year has passed for some of the placements, or buy links to the site itself to increase its inbound side.

Content and subject matter

Old content from a dropped domain is never restored. There are two reasons: copyright claims and a separate category of risks when the domain previously belonged to a government organization. The author's network includes a dropped domain of a regional health ministry website—the subject matter was adapted to the region, but not a single old page was brought back.

Instead, they create a hybrid: some materials on the old subject, some on the new one, and some at the intersection. Changing the subject through content works if the change is not abrupt. The network's subject areas are kept broad—games, IT, business, mobile apps: almost any project can fit under such umbrellas.

Looking for a topical dropped domain for a specific niche is almost useless today—finding a domain with decent link equity and the right subject is practically impossible. You should focus on the strength of the link profile and build out the subject matter through content.

Footprints that give away a network

• Shared IP. The most obvious trace. Several links from different sites on the same address work roughly like links from one domain. You should also check the subnets: if all the sites are on the same subnet, everything is on one server.
• Interlinking within the network. The basic rule is not to link your own sites to one another. Schemes involving an intermediary are best avoided as well.
• Sitewide links. The technology is considered risky: sitewide links are neither sold nor placed, except for natural ones such as a topic author's link in a WordPress footer.

Cloudflare, incidentally, is used by almost everyone—according to the author's estimate, by around ninety-five percent of networks. Different IPs and different subnets through it do not break the scheme.

How to build up the network itself

Only with normal links: guest posts, marketplace placements, and outreach. Running Xrumer and GSA blasts across network sites risks driving the network into negative territory, especially on public lists. More details on how link tiers and their safe proportions work, — in a separate breakdown.

As for updating content: experiments have shown that a year without updates does not lead to a noticeable decline in performance. Relevance decreases over time, but there is no catastrophe.

Time frames to expect

An individual link takes from two weeks to three months to take effect, most often one to two months—with the caveat that the page containing the link must also be indexed. The assessment is clouded by older links: some previously purchased links reach their full effect precisely when you place new ones.

The planning horizon depends on competition:

• competitive subject area—from a year, if you are not prepared to spend several million per month;
• low competition—from three months;
• gray-hat subject areas work differently: many sites are launched at once, some take off, some fall, and everything depends on quantity.

It is also worth remembering the domain age and documents factor. For a new site that is not on a dropped domain, there is little point expecting much in a competitive niche before a year has passed.

Controversial issues where people most often make mistakes

Disavowing links. In thirteen years of practice, disavow has helped only a handful of times, the last being in 2016, when the site had a manual penalty specifically for its link profile. At present, no effect from the file can be seen. The sensible approach is to disavow only obvious junk—a site with nofollow, zero metrics, and no traffic—and invest your efforts in new links. Google is more likely to ignore low-quality links than count them against you.

Competitor spam. A young site can be brought down with a massive volume of junk links. This no longer works with a trusted site that has traffic.

Monitoring placed links. Linkbox Pro checks whether the link exists, its type, and whether the page is indexed; the dashboard shows the share that has been indexed. Checktrust has a similar tool in beta. Monitoring indexing is essential: Google has had problems with it in recent years, especially when the referring site is unpopular and few people link to the page containing the link.

Competitors' hidden links. Studying them is almost useless. In any competitive niche, link networks and satellite sites are hidden, and what will you learn after spending a week analyzing them? That you need lots of links. Practitioners who work with networks put the conclusion more bluntly: instead of studying someone else's network, build your own.

Link echo. The skeptical position: a removed link continues to have an effect only until the index is updated—that means weeks, not half a year. The myth of a redirect that supposedly passes weight forever is also debunked: when the redirect is disconnected, the referring site is restored in the index, while the recipient loses what it received.

The link is in Ahrefs, but not in Search Console. Weight may still be passed. The Search Console sample is poorly representative, as Google representatives have also said. What matters is whether the page is in the index and whether Google can see its content, not whether it appears in the report.

Social media. If there is any effect, it is very weak. Links from there are almost always closed or go through redirects, while social networks have incomparably more outbound than inbound links.

Cheap network links with inflated metrics. Five to seven years ago, sites costing three to five dollars really did work. Now they almost never do. Sellers often do not show the URL in advance, and such referring sites burn out quickly: as soon as outbound links outnumber inbound ones, the site stops passing anything. Indian and Pakistani blogs with high DA and DR belong to the same category—in the best case, the effect will be short-lived.

What works best. Links from the homepages of powerful topical referring sites, obtained through outreach—such sites usually do not sell links. Next come network placements from a contractor you can trust. The simple sign of good work is that the links remain for years rather than disappearing after six months.

When links do not help at all. If a site has been built up with links for two years and is not growing, the problem is almost certainly not the link profile. A comprehensive audit is needed: technical factors, content, and expertise signals. In subject areas where the reader's quality of life depends on the accuracy of the material—medicine, finance, and news—links will not save you without these signals. A last resort is moving to a new domain if the current one received a black mark after an update.

It is also worth keeping the element of chance in mind: two more or less identical sites respond differently to the same links, and no one has canceled hidden filters on individual projects.

Four levels of topical relevance

A useful framework explaining why non-topical placements sometimes work while topical ones do not. Topical relevance exists at four levels:

1. the entire site;
2. the specific page;
3. the text surrounding the link;
4. the anchor itself.

Most people look only at the first level and demand that the entire referring site be dedicated to the required subject. The problem is that there are few such sites, and powerful link profiles are rare among them. Work scales at the second and third levels: a topical page can be created on almost any site.

Parasite SEO is based on the same principle: an article on a powerful aggregator domain ranks thanks to the domain's own trust, while the page content creates the topical connection. The evolution can be seen in the history of the market: first, people used bare anchors on link marketplaces, then they learned to write text around the link, and now the link lives inside a full-fledged article.

Course of action

1. Select three or four competitors with matching subject matter and geo, including one that is clearly stronger than you.
2. Compile your own referring sites by combining Ahrefs and Search Console exports, manually labeling the link type wherever Search Console does not show it.
3. Put the referring sites into a table and sort them by the ratio of inbound to outbound domains, rather than by DR.
4. Build an anchor cloud with percentages and find imbalances—both over-optimization and shortages for important keywords.
5. Investigate competitors' referring sites through pricing services, calculate the average cost and total budget, then divide it into monthly amounts.
6. Select pages that are already close to the top and choose anchors for them.
7. Place the links, recording the dates, and measure the effect no earlier than two weeks later, keeping the inertia of the old link profile in mind.
8. Keep track of the indexing and presence of placed links through Linkbox or an equivalent.