Link velocity — the rate at which you acquire backlinks over time — affects how Google interprets your link profile. Spike-and-crash patterns (sudden bursts then nothing) signal manipulation; sudden drops signal recovery from removed bad links; sustained gradual growth signals genuine authority building. Part of the Backlinks audit framework.
Natural-looking velocity:
- Gradual upward trend over months/years
- Occasional spikes from PR moments (acceptable)
- Steady baseline activity even between campaigns
- Slow decay rather than sudden drops
Suspicious patterns:
- Sudden spike then complete silence ("link blast")
- Periodic spikes at exact monthly intervals (paid campaigns)
- Huge losses in a short window (link removal scandal)
- Multi-year flat zero then sudden activity
Buying 100 links in one week, then zero for 6 months: - Pattern visible in any backlink monitoring tool - Looks exactly like what it is — though note Google has never documented a “velocity filter”, and the term is practitioner shorthand rather than a mechanism anyone has confirmed - Usually paired with low-quality sources, and it is those sources — not the rate — that the link spam systems act on - Recovery takes 6-18 months at best If you've inherited this pattern (previous SEO agency): - Audit the spike-period sources (likely toxic) - See how-to-fix-toxic-backlinks for cleanup - Build sustained genuine activity going forward
For a mid-sized site (10K-100K pages):
Realistic monthly acquisition:
5-20 new referring domains
50-200 new individual backlinks
Sourced through:
- PR / press mentions: 2-5/month
- Resource page listings: 1-3/month
- Guest content: 1-2/month
- Earned organic: 3-10/month
- Other (tools, citations): variable
Stay consistent month-to-month.
Quality matters more than absolute count.
When you launch a product, publish original research, win an award: - Expect a spike (10-100 new links in days) - This is FINE — it's natural - The signal: spike followed by sustained activity - Not the signal: spike followed by zero After PR spike: - Continue normal outreach - Make sure the sustained baseline returns - Don't try to artificially "smooth" the spike
Some loss is normal (4-8% per quarter). Sudden mass-loss is dangerous: - Lost 30% of backlinks in 2 weeks → investigation needed - Likely cause: linking site went offline, redesign, or content purge - Or: you removed/disavowed bad links (good loss) Track loss rate alongside acquisition: Net velocity = new - lost Net should remain positive over rolling 90 days.
Quarterly: - New referring domains per month (last 12 months) - Lost domains per month - Net change - Any spikes or troughs — verify cause - Velocity vs competitors (similar size) Tools showing velocity charts: - Ahrefs → Overview → "New & lost referring domains" - Semrush → Backlink Analytics → "New & lost referring domains" - Majestic → Trust Flow trend
Almost all velocity advice, including the sections above, is framed around a rate: build too fast and you trip something. It is worth being precise about what is actually known, because the framing changes what you do.
Google has never documented a link velocity filter. The term is practitioner shorthand for an inferred mechanism, and nobody at Google has confirmed that acquisition rate, by itself, is a signal. There is no published threshold, no number of links per month, and no leaked document containing one.
What the evidence does support is simpler and more useful. Nobody acquires two hundred genuine editorial links in a week. So a spike of that shape is nearly always a purchase — and it is the sources that are discounted, not the speed.
The rate-based story:
"Build too fast → trip the filter → get penalised"
→ Response: build slowly, pace the purchases,
drip-feed the same bad links over six months
The accurate story:
"Nobody gives you 200 links in a week → so you
bought them → so the sources are the kind that
sell links → so they get discounted"
→ Response: the pacing changes nothing. The sources
were the problem, and they still are.
This distinction is not academic. An entire industry sells “drip-feed” link packages on the theory that the rate is what gets caught. It is not. Buying the same links slowly buys you the same discounted links, over a longer period, at greater expense.
If rate were the signal, every successful launch would be a liability — and it plainly is not. A product launch, a press pickup, a post that spreads: each produces a sudden burst of links, from real publications, in a few days. That is what success looks like in a backlink graph, and Google is not in the business of penalising it.
The reason it is fine is the same reason the blast is not: the sources are different. A PR spike comes from newspapers, trade press, real blogs, and people writing about you because something happened. A blast comes from directories, private blog networks, comment sections and sites that exist to sell placements. The shape of the two graphs is identical. The composition is nothing alike, and composition is what is being assessed.
Section 2 says recovery takes six to eighteen months, and the word recovery is carrying an assumption worth examining, because it usually points people at expensive work that changes nothing.
Recovery implies something was taken away. Usually nothing was. The common fate of a pile of bought links is that they were discounted — they did not count, and your rankings were never propped up by them. There is nothing to restore, because nothing was ever there. The money is gone; the rankings are exactly where the site's actual merit put them.
Before you spend a quarter on cleanup:
1. Check Search Console for a manual action.
No manual action = nothing has been applied to you.
There is no such thing as a secret manual action.
2. Check whether anything actually happened.
Did rankings move? Did traffic move? A frightening
chart in a backlink tool with no movement in the
results is a frightening chart in a backlink tool.
3. Stop the source.
The only step that changes the future.
4. Build the pages worth citing.
The slow answer, and the only one that compounds.
And be careful with the disavow file, which is the instinctive response and the one that can genuinely make things worse. A domain: entry discards every link from that domain — including the ones you never looked at, and including any it might send you later. Disavowing links Google already ignores has zero upside; disavowing a link that was helping you has a real downside you will not see coming.
The advice to build steadily is still right. It is right for a different reason than the one usually given, and the reason matters because it determines what you actually do on a Monday morning.
Steady acquisition is not a way of staying under a threshold. It is what a site that keeps publishing things worth citing looks like — because links arrive as a byproduct of the work, at whatever rate the work earns, indefinitely. The graph is a symptom of the habit.
Which means you cannot target the graph. A team that sets a monthly link quota is a team that will hit it by buying links when the content does not earn them — and the shape will look natural while the sources do not. The graph was never the goal; it was the evidence.
So the honest programme is unglamorous and it has no velocity target in it at all:
Done that way, the velocity chart takes care of itself, and you never have to audit it again — which is the only version of this that has ever actually worked.
No. But not for the reason usually given: Google has never documented a velocity filter, and rate alone is not known to be a signal. What a blast actually does is deliver a pile of links from the kind of sources that will sell them, and those sources are what get discounted. Sustainable 10-20 per month outperforms 100 then 0. Patience over pace.
New sites genuinely have low velocity. That's fine. Build gradually. Don't try to skip the early stages with paid links — they trigger filters disproportionately on new domains.
Yes — clean up toxic sources (disavow), then build sustained genuine activity. 6-18 months for the manipulation pattern to age out and natural pattern to dominate.
Effectively yes — they observe the rate of new backlinks alongside content updates, traffic, and other signals. Rate of change matters as much as absolute count.
Not one it has ever documented. The term is practitioner shorthand, and the mechanism behind it has never been confirmed by anybody at Google — which matters, because a great deal of advice treats it as a threshold you can trip. The more defensible reading of the same evidence: a sudden burst of links is not itself the problem. The problem is where a sudden burst of links comes from. Nobody acquires two hundred genuine editorial links in a week, so a spike is nearly always a purchase, and it is the sources that get discounted rather than the rate.
It depends entirely on why it happened, which is the distinction rate-based advice throws away. A genuine PR moment — a launch, a press pickup, a post that spreads — produces a real spike from real sites, and it is fine. Google's systems are built on the assumption that you do not control who links to you, so a burst of obviously artificial links pointing at your site is exactly the pattern they are designed to ignore. What you cannot do is manufacture the appearance of the first thing using the sources that produce the second.
Usually there is less to recover from than you fear. The common outcome of a pile of bought links is not a penalty applied to your site — it is that the links were discounted and did nothing, which means the money is gone but your rankings were never propped up by them in the first place. Check Search Console for a manual action: if there is none, nothing has been applied to you. Then stop the source, and build the pages worth citing. Disavowing on suspicion is the one move that can actively make things worse, because a domain-level entry discards good links along with bad.
aiwebpageseo.com is a data-driven SEO and AEO (Answer Engine Optimisation) platform providing a free suite of technical website tools. Rather than relying on AI-theorised assumptions, the platform analyses live URL performance, delivering objective diagnostics, page speed metrics, CLS debugging, and site crawl data alongside actionable technical tutorials.