- Measure your own referring-domain count per month for the past year, and ramp new link activity in steps that sit close to that baseline.
- Judge a link burst by its shape and sources rather than its size: check that it's irregular, fades after its news hook, and comes from sites with real traffic and indexed pages.
- Log two dates for every link, the day it was earned and the day its page was indexed, so you don't mistake indexing batches for a sudden spike.
Link velocity sounds like a speed limit. Find the number, stay under it, and your link profile is safe. Forum threads and SEO blogs will happily offer you one. Google has never published a number, and the reason it hasn’t is more useful than any number would be.
Most people asking about link velocity are really worried about a spike: a month where referring domains jump, or a campaign that suddenly lands forty links in two weeks. The fear is that a sudden jump looks like manipulation. Sometimes it does. Often it’s what happens when something good happens to a business: a product launch, a study other writers cite, a newsletter that features you. The question that matters is not how fast the links arrived. It’s whether the pattern looks like something real people did.
Here’s how to answer that, starting with what the term actually measures.
What link velocity actually measures
Link velocity is the rate at which a site gains new links, usually counted as new referring domains per week or per month. It’s an SEO-industry term, not a Google one. Google doesn’t publish a velocity score and doesn’t rank sites by how quickly they pick up links.
You can only read velocity against a baseline. Forty new domains in a month is alarming for a site that normally earns two. For a site that has added thirty a month for two years, it’s an ordinary month. Anomaly detection, the general idea behind most spam filtering, looks for departures from a baseline, and every site has a different baseline. A rate that looks extreme on one domain is unremarkable on another.
That’s the first practical point. Velocity isn’t a property of a single link. It’s a property of a history, and you only know your history if you measure it. Most teams that panic about link velocity have never looked at their own referring-domain count over the past year.
Google has no speed limit, only a list of methods
Google’s guidance on links is about how links are obtained, not how fast they arrive. Its Search Essentials documentation lists the link schemes it considers spam: buying or selling links that pass PageRank, excessive link exchanges, automated link-building programs, and large-scale article or guest-post campaigns that use keyword-rich anchor text. Read that list again. Every item is defined by method or intent. None is defined by a weekly count.
So a fast campaign can be fine, and a slow one can be a clear violation. A patient program that buys placements from a single link network is still a link scheme. A sudden burst of links from journalists who chose to cite your research is not. Speed is at most a symptom. The method is the violation.
The enforcement history points the same way. Penguin, which first rolled out in 2012, targeted manipulative link patterns, and Google folded it into its core ranking systems in 2016. Much of the current work runs through SpamBrain, Google’s AI-based spam system. Often the outcome is quiet: the link simply stops counting, and nobody is penalized. Manual actions, where a human reviewer applies a penalty, are reserved for serious cases and appear in Search Console as “Unnatural links to your site.”
The practical lesson is to stop designing around a speed limit you hope exists. Design around the question a reviewer would ask: does each link exist for a reason a real person would recognize?
What Google’s systems actually read in a link profile
Natural link profiles are messy in consistent ways. Four things differ between earned and manufactured growth, and each comes from how the links were created, not how many arrived.
Who is linking. Earned links come from people with no relationship to you: a blogger who read your post, a forum user who needed a statistic, a journalist writing about the problem. Manufactured campaigns tend to draw from a narrow set of source types, often the same network of sites, the same template, or the same owner.
What the anchor says. Real writers link with whatever fits their sentence. Expect a mix of your brand name, a bare URL, “this calculator,” “the team at,” and descriptive phrases. Uniform exact-match keyword anchors are among the most common giveaways, and it’s the uniformity that stands out, not the keyword itself.
Where the link sits. Editors add links inside the body of a paragraph where the reference makes sense. Footers, sidebars, and author-bio blocks that repeat across many domains look like a template, because they are one.
What happens after. Earned attention decays. A link from a news hook gets plenty of traffic for a week, then the tail thins out as writers keep citing the piece for months. Manufactured placements tend to look either flat or cut off the moment the campaign stops, with no tail at all.
The table below puts these signals side by side.
| Signal | Typically earned | Typically manufactured |
|---|---|---|
| Timing | Clusters around a launch, study, or press hook | Spread evenly, or packed into one window with no hook |
| Source sites | Unrelated sites, many with real audiences | Shared network, template, or owner; little organic traffic |
| Anchor text | Brand name, bare URL, “this tool,” descriptive phrases | One exact-match keyword phrase, repeated |
| Placement | Inside relevant paragraphs | Footer, sidebar, author bio, or one sitewide link |
| After publication | Decays over weeks, with a thin long tail | Flat, or stops abruptly when the campaign ends |
| Clicks | Referral visits from readers | Few or none |
A fast burst that’s fine, and a slow one that isn’t
Consider a hypothetical invoicing app. Its team publishes a free late-payment fee calculator in March. A payments newsletter features it on a Tuesday, and that’s one link. A finance blog cites it in a roundup, two more. Over the next three weeks, new referring domains come in at something like 6, 22, and 9, then taper to a handful a week for the rest of the spring. The linking sites have nothing to do with each other. The anchors vary: the product name, “this calculator,” a bare URL, a sentence that happens to contain “late payment fees.” The burst lines up with a launch and a news hook. That’s a real burst.
Now the same app, the same month. Forty guest posts go live over two weeks, on sites in pet insurance, cryptocurrency, travel, and home improvement. Most of those sites were created within the past year and get almost no organic traffic. Each post runs a few hundred words, each links once in the body, and each uses the anchor “invoicing software.” The posts follow one template. Counted by week, it’s twenty and twenty.
The totals are similar, and so is the rate. Neither number is the problem. The first burst looks like what happens when people find something useful. The second looks like what happens when someone orders it. What separates them is everything around the count: who linked, why, with what words, and what happened afterward. Guest posting isn’t the issue here. Unrelated sites, a shared template, and identical anchors are. If the relevance logic is unfamiliar, our piece on why relevance beats volume explains why the second group was worthless even before the pattern gave it away.
Patterns that make a spike look engineered
These are the tells that matter most in practice. None requires a special tool to spot once you know what to look for.
- A rigid rhythm. Identical counts on a fixed schedule are the most recognizable pattern in purchased links, because a schedule is what produced them.
- Source clusters. Open a backlink export and sort by first-seen date. A cluster of new domains with near-identical layouts, created in the same few months, all linking to the same URL, is a shape you can see at a glance. Your monitoring tool makes it visible, and search engines don’t need your export to notice it.
- Links from pages no one visits. A link on a page with no traffic, no indexing, and no internal links pointing to it exists only for the link. It passes no value until its page is indexed, and a page that’s never indexed never passes any.
- Repetition on one domain. One site with a sitewide link, or the same author bio pasted onto thirty posts, looks like a template regardless of timing. A single domain repeating a link pattern is a pattern on its own, even if the rest of your profile looks clean.
- A spike with no brand trail. Real attention usually leaves other traces. People search for the brand, mention it without linking, and click through from the pages that link to you. A jump in referring domains with flat brand search and no referral clicks is harder to explain as earned. It looks placed.
Regular beats fast: fifteen links every Monday for three months looks engineered even when the total is modest. Real acquisition is lumpy, because real opportunities don't arrive on a calendar.
If you find a cluster you didn’t create, the disavow tool exists for exactly that situation. Treat it as a last resort for links you can’t get removed, not a routine cleanup. Disavowing links that were fine doesn’t help, and it’s easy to do by mistake.
How real sites earn links in bursts
Earned bursts have a recognizable shape, and it’s worth knowing before you launch anything. The first wave comes from the sites with the most reach: a newsletter, a major industry publication, a popular forum thread, a social account with a real following. A second wave follows, made up of writers who saw the first wave and decided to cite the same source in their own pieces. That wave is slower and wider. Then a long tail arrives as the piece gets pulled into roundups, reading lists, and later articles that need a statistic. A healthy burst is steep at the start and thins gradually over weeks to months, depending on how evergreen the subject is.
Most bursts that real businesses earn come from a short list of triggers:
- An original dataset or study. Writers cite numbers they can’t get anywhere else. A survey of 800 freelancers with one clear finding gets referenced for years in a way a generic opinion post never does.
- A free tool that fills a specific gap. Calculators, checkers, and templates earn links because people reference them while doing a task. The link goes where the work happens.
- A product launch with a real story. A launch earns links when there’s something to say about it: a problem it solves, a number that moved, a decision behind it. “We launched” by itself rarely gets cited.
- Press or podcast coverage. Interviews often land with show notes or a write-up that links back. Those links arrive in clusters tied to a date you can trace.
- A community moment. A detailed thread or a genuinely useful answer that other people pick up and pass along.
The burst is earned by the thing. The links are earned by the people who cite it. Planned bursts are legitimate too. A digital PR campaign built around a study is designed to produce exactly this shape, and it runs on the same mechanism as the calculator launch: the team prepares the hook, pitches the people who cover the topic, and accepts that the timing is partly out of their hands.
Pacing link building so it doesn’t read as a pattern
If you run outreach or placement programs, you control the pace of the parts you can plan. This process keeps your numbers tied to real opportunities.
- Pull a baseline first. Export twelve months of referring-domain history from a monitoring tool. Search Console’s Links report shows totals and top sources, but no dated timeline of new links, so it can’t show you how your velocity has moved. Find your normal monthly range before you plan anything.
- Set targets relative to that baseline. If your site adds three to five referring domains a month on its own, a plan that adds forty in one month is a step change nobody can explain. Ramp in increments you could justify to a skeptical editor, and raise the ceiling only after the previous step has settled.
- Break the rhythm on purpose. Schedule placements in irregular batches, and let some weeks be quiet. A month with twelve new links followed by one with three looks like work that follows real opportunities, because that’s what it is.
- Expect indexing to reshape your numbers. Links become visible in batches as the pages hosting them get crawled. If a month of placements is indexed in the same week, your dashboard shows a spike even though the links were earned gradually. Why backlinks aren’t moving rankings covers why that lag happens and how to check it.
- Give the links something to point to. A study, a tool, or a resource creates natural reasons for placements to appear over time. Links aimed at a page with no reason to exist are harder to justify, and harder to earn.
- Review monthly. Check the exact-match share in your anchors, look for new source clusters, and confirm the new links send readers. Those three checks catch most of the problems described above.
Track two dates per link: the day it was placed or earned, and the day its page was indexed. The first tells you how fast you're working. The second tells you how fast Google sees the work. Teams misread their own spikes when they mix the two up.
Running this by hand across dozens of sites takes real effort, which is why OptiLink builds the placement and indexing work into one workflow. Each link is hand-placed in content where it fits, checked for dofollow status, pushed to get indexed, and tracked live. The pace of the profile then comes from what actually got placed, not from a schedule someone set in advance.
What to take from this
Link velocity does have a too-fast, but it’s defined by pattern and by your own history, not by a number Google published. A burst is fine when it’s lumpy, comes from sites with reasons to link, uses varied and natural anchor text, and fades the way real attention does. It’s a problem when the rhythm is rigid, the sources cluster together, the anchors repeat, and nothing else about the business moved.
The useful habit is measurement. Know your monthly baseline, watch for clusters and repetition, and keep the day a link was earned separate from the day it was indexed. Then build a program where speed is a byproduct of real placements rather than a target you aim for.
That’s the logic behind OptiLink. Every placement goes into content where the link genuinely fits, every page is checked for indexing, and every link is tracked live, so the pace of your profile reflects real work instead of a calendar. If you want to see how that looks for your own site, start with how OptiLink works.
Keep reading
Anchor text optimization without tripping a filter
A hundred random links can do less than five relevant ones. Here's what makes a backlink pass real ranking value in 2026 — and what makes it worthless or worse.
Sofia Meyer
September 24, 20262 min readBroken link building, step by step: find, fix, and pitch
A hundred random links can do less than five relevant ones. Here's what makes a backlink pass real ranking value in 2026 — and what makes it worthless or worse.
Sofia Meyer
October 9, 20262 min readCompetitor backlink analysis that finds real opportunities
A hundred random links can do less than five relevant ones. Here's what makes a backlink pass real ranking value in 2026 — and what makes it worthless or worse.
Sofia Meyer
September 26, 20262 min readDofollow vs nofollow in 2026: what still passes value
The link attributes changed, Google's treatment changed, and a lot of advice is out of date. Here's what dofollow, nofollow, sponsored and ugc actually do now.
The OptiLink Team
October 10, 20262 min read