A profile is a shape, not a score
Every tool that shows you backlinks also shows a number — DA, DR, Trust Flow, Authority Score. Those are third-party estimates built to be comparable at a glance, and they're the least informative thing on the screen. Two sites with identical DR 40 can have completely different profiles, one of which survives the next core update.
What an auditor reads instead is the distribution: where links come from, how varied the sources are, what words they're wrapped in, when they arrived, and whether the pattern has a plausible cause in the real world. A profile is a story about how a business became known. Real stories are uneven. Manufactured ones are regular in ways that are easy to spot.
One thing upfront: the ranges below are practitioner heuristics, from looking at a lot of profiles and knowing which ones later had problems. Google publishes no thresholds. Anyone quoting an exact safe percentage is dressing up the same guesswork.
The six ratios, and where healthy sits
Run these in order. Any backlink tool gives you all six, free tiers included, and Search Console's links report covers the first two adequately.
| Ratio | What healthy looks like | What a failure means |
|---|---|---|
| 1. Referring domains : total backlinks | Roughly 1:5 to 1:50 for most sites. Older sites and news publishers sit higher. | 1:500 and up means sitewide footer, widget or template links inflating the count. Ignore the big number. |
| 2. Branded and bare-URL anchor share | Usually the majority — brand name, domain name, "here", the page title. | If exact-match commercial anchors are more than roughly 10–15% of anchors, no natural process produced that. |
| 3. Dofollow : nofollow | A real profile carries a meaningful nofollowed share — social, forums, big media, Wikipedia. | Near 100% dofollow is the clearest purchase signal there is. Nobody earns only followed links. |
| 4. New referring domains per month | Uneven, and correlated with something you did — a launch, a study, press coverage. | Vertical spikes with no cause, or perfectly identical monthly counts. Both are subscriptions. |
| 5. Topically relevant domain share | A clear majority from your sector or adjacent to it, plus local and general press. | A dental clinic with links from crypto, casino and essay-writing sites is holding somebody else's inventory. |
| 6. Geographic and language mix | Broadly matches where you sell. An Indian B2B firm should skew Indian, with some global trade press. | Heavy concentration in unrelated markets and languages is the signature of a cheap link package. |
The ten-minute audit
This is the actual sequence, in the order that finds problems fastest. It works on your own site and just as well on a competitor's, which is the more interesting use.
- Pull referring domains, not links. Then ignore whatever authority metric your tool shows and just read the domain names. Thirty seconds of that catches most of what an algorithm would flag.
- Open the anchor text report and sort by frequency. If a commercial phrase sits above your brand name, you've found the problem already.
- Check the dofollow split. Anything above roughly 90% followed deserves an explanation.
- Look at the referring domain graph over time. Shape, not slope. Steps mean campaigns; smooth vertical climbs mean purchases.
- Sample twenty referring pages at random and actually visit them. The step everyone skips and the one that settles it. Does the page look like something a person would read? Is your link surrounded by six unrelated commercial links?
- Check where the links point. Homepage-only distribution on a fifty-page site means the links were bought for a number, not a ranking.
What a bought profile looks like from the outside
Once you've read a few, purchased profiles become obvious. They share tells because a small number of suppliers produce them from the same playbook.
- Guest posts on sites that publish nothing else anyone reads. Twelve unrelated industries in one blog category, every post around 800 words with two outbound commercial links.
- Identical anchor text across dozens of domains. Real writers phrase things differently. Order forms don't.
- Links appearing in a burst, then nothing. Somebody bought a package in March and let it lapse.
- Domains with high authority scores and no traffic. Expired domains rebuilt to sell links. The metric survived; the audience didn't.
- Referring sites whose own outbound links are 90% commercial. Opening a referring site's link report is the fastest way to identify a network.
- Footer links across an entire template, usually a web developer's client sites all pointing at each other.
When a disavow is justified — and when it backfires
The disavow tool tells Google to ignore specific links. It's useful in a narrow set of cases and harmful as routine hygiene, which is how a lot of tools market it.
Google's position has been consistent for years: most sites never need it, because the systems already discount links they don't trust. It exists mainly for cases where you know something bad was built and Google might reasonably hold you responsible.
Justified
- You have a manual action for unnatural links in Search Console. The clearest case, and it usually needs a removal effort alongside the file.
- You know links were bought — by you or a previous agency — and can identify them specifically.
- A penalty-era legacy on an old domain, where thousands of directory or PBN links were built years ago and you're cleaning up before a relaunch.
Not justified
- A tool gave some links a red "toxic" score. That's the vendor's opinion, not Google's, and it's frequently wrong about legitimate sites.
- Rankings dropped and you don't know why. Core updates, competitor improvement and intent shifts explain far more drops than links do. Diagnose first.
- Someone pointed spam at you. Google's systems assume this happens and largely ignore it — see why links only break ties.
What you're actually trying to build
The target isn't a number. It's a profile a sceptical human would find plausible reading it top to bottom — links from places your customers, suppliers and industry actually read, acquired at a rate that matches what your business has been doing, described in the words people naturally use about you.
For a mid-size Indian business that usually means industry associations and trade press, the business media in your city, suppliers and partners who list you, a piece of original data people cite because it's the only source, and a slow drip of citations from people you've never spoken to. Slower than buying, and the only version that keeps its value.
For what that costs properly, link building costs breaks the numbers down — including why the cheap tier can't produce this profile at that price.