Link Building Guide

We sell links, which is worth stating before you read another word. SerpInsight places link insertions (also called niche edits) into articles that already rank, and that commercial interest sits behind every opinion on this page. We’ve tried to write it so that someone who never buys a single placement from us still leaves better informed than they arrived.

That means being blunt in a few places. Several of the tactics below are a poor trade for most sites. One of the most reliable things you can do costs nothing and needs no outreach at all. And if your site is genuinely new, the honest answer is that links are not the thing standing between you and traffic yet.

How to use this link building guide

Read it top to bottom if the subject is new to you. The sections build on each other, and the tactics only make sense once you understand what separates a link that moves rankings from one that does nothing. If you already run campaigns, jump straight to whichever section names your actual problem.

Every section ends with a pointer to a deeper post on that specific question, so treat this page as the map rather than the whole territory. The detail lives in the cluster; the shape of the thing lives here.

• What link building actually is — PageRank, why links still count, and what Penguin changed
• What makes a link valuable — the six attributes that decide it, and how to check each one
• How to build backlinks: the main tactics, honestly assessed
• Link building strategies compared side by side
• Tactics we would avoid, and why white hat link building wins on cost
• What anchor text should you use?
• How many links do you need, and how fast?
• How to measure link building without fooling yourself
• A backlink acquisition plan for where you actually are
• Frequently asked questions

What link building actually is

Strip away twenty years of jargon and the mechanic is simple. A link from one page to another is a vote, and search engines have counted votes since 1998, when PageRank proposed that a page linked to by many important pages is probably itself important. That recursive idea — importance flows from important sources — is still the skeleton underneath modern ranking systems, even though it has been buried under hundreds of other signals since.

What changed is how the votes are weighed. Early PageRank counted links fairly mechanically, which is why the first decade of SEO was largely an arms race in volume. Directories, article farms, reciprocal link pages, footer links sold by the thousand. It worked, embarrassingly well, until it didn’t.

The Penguin update in 2012 was the turning point. Google started devaluing and in some cases actively penalising unnatural link patterns, and a lot of businesses that had bought their way to page one discovered how quickly that position could be taken back. Penguin was folded into the core algorithm in 2016 and its behaviour shifted: rather than penalising a site outright, the system now mostly ignores links it doesn’t trust. That’s a meaningful difference. The common outcome of a bad link campaign in 2026 isn’t a manual action — it’s spending money on links that were quietly discounted and never counted at all.

So do links still matter? Yes, and anyone selling you a story about links being dead has usually replaced them with something less measurable. Google’s own documentation still treats links as a signal, its spam policies still single out link schemes for specific attention, and if links carried no weight nobody would bother writing policies about them. The more useful framing is that links have become one input among several, and their weight varies enormously by query type. For a competitive commercial term where five sites all have excellent content, referring domains often decide the order. For a long-tail informational query with weak competition, a well-written page can rank with almost none.

The other shift worth naming is E-E-A-T — experience, expertise, authoritativeness and trust. Links contribute to how those qualities are assessed, but they contribute as evidence rather than as currency. A link from a respected publication in your field says something about you. A link from a site that exists only to sell links says something too, and it isn’t flattering.

What makes a link valuable

Not all links are worth the same, and the gap between the best and the worst is far wider than most pricing suggests. A link from a well-read industry publication that sends you actual readers might be worth fifty links from sites nobody visits. Yet the fifty often cost more in total. That’s the arbitrage most link buyers get wrong.

Six attributes decide most of a link’s value. Topical relevance comes first and is the one people most often trade away for a bigger number. A link from a site that covers your subject carries context; a link from a general-purpose blog that publishes about roofing on Monday and cryptocurrency on Tuesday carries very little, no matter what its Domain Rating says.

Authority comes second, and it’s worth separating domain-level from page-level. Domain Rating, Domain Authority and their equivalents are third-party estimates of a site’s overall link strength — useful for sorting, useless as a target in isolation. Page-level authority matters more than most people account for, because a link on a page that itself has no links pointing at it passes considerably less than a link on a page that’s well cited.

Real traffic is the check that catches the largest share of bad inventory. A site can have a DR of 65 and receive four hundred visits a month, which usually means its authority was built rather than earned. We require every host site to have at least 1,000 monthly organic visits before we’ll place on it, and we’d suggest a similar floor whether you’re buying or earning.

Then placement, anchor and link attribute. A link in the third paragraph of the main body, surrounded by relevant prose, is worth more than the same link in an author bio or a footer. The anchor text tells search engines what the destination is about. And a followed link passes signal in a way a rel="nofollow" or rel="sponsored" link generally does not — though nofollowed links from genuinely prominent places still have value for referral traffic and for how they look in a natural profile.

If you want the longer version of this assessment, including the specific red flags that get a site rejected in our vetting, we’ve written it up in a dedicated post on how to tell a link worth having from expensive filler. And if you’re wondering what these attributes translate to in money, the pricing patterns are broken down in what backlinks actually cost across the market.

How to build backlinks: the main tactics, honestly assessed

Eight tactics account for nearly all legitimate link acquisition. What follows is a description of each with a candid read on effort, risk and whether it scales. Three of them are excellent for most sites, three are situational, and two are widely oversold.

Link insertions (niche edits)

You find an article that already exists, already ranks and already gets readers, then arrange for your link to be added inside it where it genuinely fits. No new content is created. The host page keeps its existing authority and indexation, which is the entire appeal — you’re not waiting months for a fresh post to earn standing.

Effort is moderate if you do it yourself: roughly four to six hours per placed link once you include prospecting, vetting, outreach and follow-up, and you’ll typically need to contact eight to fifteen sites for each acceptance. Risk is moderate and depends almost entirely on discipline. A relevant insertion into a paragraph where the link adds something is very hard to distinguish from an editorial citation. A link jammed into an unrelated sentence on a site that sells fifty of them a month is not. It scales reasonably well because the inventory of existing articles is enormous. We explain the mechanics properly in our breakdown of what a link insertion is and how placements are made, with concrete before-and-after paragraphs in these worked placement examples.

Guest posting

You write an article for someone else’s site and include a link to yours, usually in the body and sometimes in the author bio. It’s the most established paid-adjacent tactic and still works when the host site is real and the article is genuinely good.

Effort is high — you’re producing a full piece of content per link, on top of the outreach. Risk is moderate: guest posting at scale with thin articles on sites that exist to host guest posts is exactly the pattern spam systems look for. It scales poorly for that reason, since the quality ceiling drops fast as volume rises. The upside is control. You choose the angle, the surrounding context and the anchor. If you’re weighing the two approaches against each other, this comparison of niche edits and guest posts works through when each makes sense, and we keep a set of outreach templates that get replies for anyone doing it in-house.

Broken link building

You find links pointing at pages that no longer exist, create or identify a suitable replacement on your own site, and tell the linking site about it. The pitch has a built-in reason to say yes, which is rare in outreach: you’re fixing something on their site.

Effort is high per link and heavily front-loaded into prospecting. Risk is close to zero — nothing about it violates any guideline. It scales badly, because the supply of relevant broken links in any given niche is finite and you’ll exhaust the good ones. Reply rates are usually better than cold outreach but the conversion from reply to live link is slower than people expect. It’s a good tactic to run continuously in the background rather than as a campaign. The full method, including the tooling, is in our walkthrough of broken link building.

Digital PR and linkable assets

You create something worth citing — original research, a genuinely useful free tool, a data set nobody else has published — and then promote it to journalists and bloggers who cover that beat. When it works, it produces the best links available: editorial, unpaid, from publications that don’t sell placements at any price.

Effort is very high and the cost is real, both in production and in outreach time. Risk is nil. It scales in bursts rather than steadily, and the variance is brutal. A campaign either lands and brings forty referring domains or lands nowhere and brings two. Most small businesses shouldn’t attempt this until they have something legitimately new to say. When you do, one good asset can outperform a year of transactional placements.

Expert quote sourcing (HARO-style)

Journalists request expert commentary through query services; you respond with a usable quote and get credited with a link. The mechanics have shifted since the original HARO wound down, but the pattern persists across several replacement platforms.

Effort is moderate and recurring — you’re monitoring queries daily and writing thoughtful responses that mostly get ignored. Risk is nil. It scales only as far as your own time and genuine expertise stretch, and this is where it falls down for most people: generic answers written to a formula are obvious to journalists who receive two hundred of them. If you have real operating experience in your field, the hit rate is respectable. If you’re outsourcing the responses, it usually isn’t worth the subscription.

Resource page links

Some sites maintain curated lists of useful links on a topic — university department pages, industry association directories, well-maintained roundups. Getting added is a matter of finding the list, confirming you belong on it, and asking.

Effort is low per attempt but the accept rate is unpredictable. Risk is nil provided you’re only pursuing genuine curated resources rather than pay-to-list directories dressed up as resource pages. It scales poorly because the good lists are few and most were abandoned in 2019. The tell for a real one: it’s been updated in the last two years and its outbound links still work.

Unlinked brand mentions

Somebody has already written about you, quoted you or named your product, and simply didn’t link. You ask them to. This is the highest-conversion outreach there is, because the editorial decision to mention you has already been made.

Effort is low. Risk is nil. It scales exactly as far as your existing mention volume, which for a new business is often zero — which is the catch. Set up a monitoring alert for your brand name, check it monthly, and send a short polite note whenever a mention appears without a link. Expect to convert a decent share of them, and expect the pool to stay small until your brand has some reach.

Internal linking

Links from your own pages to your own pages. It costs nothing, requires no outreach, needs nobody’s permission, and it’s the most neglected tactic on this list by a wide margin — because nobody sells it, so nobody markets it to you.

Internal links distribute authority around your site, tell search engines which pages you consider important, and give crawlers a route to pages buried in archives. If you have a page you care about and fewer than three other pages on your site link to it contextually, you have a free improvement available this afternoon. Do this before you spend anything externally. The specifics — anchor selection, hub structure, how to find orphaned pages — are in our guide to internal link structure that actually distributes authority.

Link building strategies compared side by side

The table below is our read on the trade-offs, based on how these tactics behave generally rather than on measured outcomes from a specific data set. Effort figures assume you’re doing the work yourself, and they vary a lot by niche — outreach in software is a different sport from outreach in local trades.

Tactics we would avoid, and why white hat link building wins on cost

The tactics below still get sold, still get bought, and still occasionally produce a short-lived ranking bump. We avoid all of them, and the reason is arithmetic rather than moralising. Each one has an expected value that turns negative once you account for the cleanup.

Private blog networks. A PBN is a set of sites, usually built on expired domains, that exist to link to clients. The economics look excellent right up until the network gets identified, at which point every link in it stops counting at once and you’ve lost the entire investment simultaneously rather than gradually. Concentrated risk is the problem, not the individual link quality.

Comment and forum spam. Nearly every comment link has carried nofollow or ugc for years, so the ranking value was largely eliminated a long time ago. What remains is a footprint that makes your profile look automated, plus the reputational cost of being the brand that spams other people’s threads.

Paid directory blasts. The packages promising two hundred directory listings for forty dollars are selling links from sites with no editorial standards, no readers and no reason for search engines to trust them. A handful of directories are genuinely worth being in — the industry association for your trade, a legitimate local chamber listing. Those you should submit to individually.

Link exchanges at scale. Reciprocal linking between two genuinely related businesses is unremarkable and happens naturally. Organised exchange schemes, three-way rotations and “link partner” spreadsheets are a different thing. They produce a detectable pattern, and the pattern is the whole point of the scheme, which makes it hard to hide.

Automated outreach at volume. Sending five thousand templated emails to scraped addresses does technically produce links, at a reply rate low enough that the maths only works if you send enormous volumes of low-quality mail. It burns your sending domain, it annoys the exact publishers you’ll want later, and the sites that accept that kind of pitch are rarely worth being on.

What anchor text should you use?

Aim for a profile that’s mostly branded and partial-match, with exact-match anchors kept to a small minority. As a working target we use roughly 40–50% branded, 25–35% partial or topic-based, 10–20% naked URLs and generic phrases, and 5–10% exact-match. Those are guides for the shape of the whole profile, not rules for individual links.

Anchor text is where most bought link profiles give themselves away. A site with fourteen referring domains, eleven of which use the exact phrase the site is trying to rank for, has published a confession. Natural citation doesn’t behave that way — people link using your brand name, the article title, a descriptive phrase they made up, or just the bare URL.

The practical approach is to audit before you add. Pull your existing anchor distribution, work out which buckets are underweight, and let that decide the anchor for your next placement rather than deciding it from the keyword you want. If exact-match is already at 20%, your next several links should be branded regardless of what would feel most useful. The ratios, the buckets and how to run the audit are covered in detail in our anchor text ratio guide.

How many links do you need, and how fast?

There’s no universal number, and any provider quoting one hasn’t looked at your competition. The usable method is comparative: pull the referring domains for the pages currently ranking in the top five for your target term, take the median rather than the average, and compare it to your page. That gap is your target, and it’s page-level, not domain-level.

Pace matters more than volume for most sites. A steady four to eight placements a month, sustained, tends to produce a healthier profile than thirty links in one month followed by silence — partly because sudden spikes look unnatural, and mostly because consistency is what compounds. The exception is a site that just published something genuinely newsworthy, where a burst is exactly what natural coverage looks like.

Speed of results is the other half of the question. Expect a lag of several weeks between a link going live and any observable movement, and expect the signal to be noisy for longer than that. We go through how to set a monthly target against your own competition in this piece on how many backlinks per month you actually need, and the question of whether placements stay live is handled in our look at how long backlinks typically last.

How to measure link building without fooling yourself

Track referring domains, not total backlinks. Track them at the page level for the pages you’re actually working on. Then track the rankings and organic traffic of those specific pages, and give the whole thing at least three months before you draw a conclusion. Everything else is secondary.

Total backlink count is the classic vanity metric. One site-wide footer link on a blog with four thousand pages will add four thousand backlinks and roughly one referring domain’s worth of value. Domain Rating is nearly as misleading when treated as an objective — it’s a third-party estimate, it moves for reasons unrelated to your work, and no search engine uses it. It’s useful for sorting prospects and useless as a KPI.

The metrics that repay attention are narrower. Referring domains to the target page. The position of that page for its primary term and for the cluster of secondary terms around it. Impressions in Search Console for the page, which often move before position does and give you an early read. Referral sessions from the linking sites, which tell you whether the placement is reaching humans as well as crawlers. And link survival — whether what you paid for is still live six months later, which is why we replace removed links free for twelve months.

On timing: the data doesn’t become readable quickly. Individual links rarely produce a visible step change; the effect shows up as a trend across a batch. Plan to look at a rolling three-month window and compare it to the previous one, rather than checking rankings on Tuesday because a link went live on Monday. That habit produces false conclusions in both directions and has caused more abandoned campaigns than any algorithm update.

A backlink acquisition plan for where you actually are

The right first move depends almost entirely on your starting position. Here are three, described honestly.

1. New site, under roughly DR 20, thin content

   Links are not your constraint. We’ll say that plainly even though it costs us business. A new site with twelve pages and no rankings has a content and coverage problem, and pointing links at pages that aren’t competitive won’t change the outcome. Spend the next quarter publishing genuinely useful pages that answer specific queries, get your internal linking right, register the handful of legitimate industry and local listings that apply to you, and build the site people would want to cite. Come back to external links when you have pages ranking somewhere in the top thirty for terms you care about — that’s the signal that authority has become the deciding factor.

2. Established site with solid content, ranking on page two

   This is where link building earns its keep. Your pages are already in the candidate pool and search engines have judged them plausible; the gap between position 14 and position 6 is frequently authority. Start by picking three to five pages that are close and commercially meaningful, not the whole site. Compare each one’s referring domains against the median of the top five results. Then run a steady programme — four to eight placements a month across those pages, mostly link insertions on relevant hosts, anchors chosen to balance your existing distribution rather than to chase the keyword. Fix internal linking to those pages in the same week, because it’s free and it compounds with everything else.

3. Competitive niche where everyone has good content and good links

   Transactional placements alone won’t win this. When the top ten all have strong content and hundreds of referring domains, the differentiator has to be something they can’t buy. That usually means original data, a free tool people bookmark, or genuine expert visibility — the digital PR and expert-quote end of the list. Keep a baseline programme of insertions running to hold parity, but accept that parity is all it buys. The move that changes your position is publishing something the industry has to cite. It’s slower, more expensive and considerably less certain, and in a niche like that it’s also the only thing that works.

Across all three, the sequencing principle holds: fix what’s free before you buy what isn’t. Internal links, indexation, intent match and content depth all cost time rather than money, and every one of them makes purchased links work harder when you get to them.

Frequently asked questions

Where to go next

If you’ve read this far, you have enough to decide whether links are your bottleneck and which tactic fits your situation. The cluster posts linked throughout this link building guide go deeper on each individual question, and they’re worth reading before you spend anything.

If you’d rather hand the work over, SerpInsight runs placements into articles that already rank, along with commissioned guest articles and relationship-led outreach. Every host site clears the same vetting bar described above, turnaround is 7–14 days per placement, and removed links get replaced free for twelve months. Our per-link pricing is published in full with no packages to decode. If you’d like a straight answer on whether links are the right spend for your site right now, tell us what you’re working on — we reply within 24 hours, and we’ll say so if the answer is no.