An e-commerce site runs a thirty-placement plan for a category page it has just launched. The list of sites is sorted by authority score and the highest-scoring ones are chosen. The same commercial keyword is used as the anchor in every article. The posts are spread out at one a day so that they "look natural". Three months later, there is no meaningful change in the page's position.
None of the steps in this plan were malicious. All four were decisions made in good faith, following common advice. What they share is this: each one optimizes not the link itself but the number shown by the tool that measures it. The score is an estimate, the schedule is an appearance, the anchor is a row in a counter. Most of the decisions that do damage in link building come not from buying cheap links but from exactly this substitution.
The one question that makes a link decision right
Google's spam policies define link spam by purpose, not method: the practice of building links whose primary purpose is to manipulate rankings. The definition does not mention a site's score, age or price.
In practice, this comes down to a single question: if search engines did not exist, would this link still be on that page? If the answer is "yes, because the article really is about the topic and readers have a reason to click", the link is in editorial territory. If the only answer you have is "we paid for it" or "we exchanged links", the site's score does not change that picture.
The test fails most clearly with links inserted into a site's code without the owner's knowledge; how this method works and why it puts both sides at risk is covered separately in what is a hacklink. For the basic framework of what a link is and how it carries value, what is a backlink is the starting point.
The question is not a moral warning but an operational filter. Google handles spam not through individual rule lists but through a system that recognizes patterns, and the scope of the policy today covers not only classic rankings but also attempts to influence generative AI answers within Search. A link set produced from a template has distinctive traces, and those traces show up not in a single placement but across the set as a whole.
What "spread the links out by day" does not solve
This request comes up in almost every project: let's spread the posts across the calendar so that a sudden pile-up doesn't look artificial. The logic is understandable, but it does not solve the problem it thinks it solves.
Real link acquisition is not evenly distributed. A product launches, data is published, a topic hits the news and you get mentioned several times in a few days; then nothing happens for months. A line that produces exactly one link a day has no counterpart in real publishing. The pattern chosen to look natural is itself an unnatural pattern.
More importantly, what the policy looks at is not the schedule but the source. Google's examples of link spam include links exchanged for money or products, excessive reciprocal link exchanges, links created by automated programs, links spread across the templates and footers of various sites, and low-value content produced solely to carry links. Slowing any of them down doesn't take it off the list.
The right approach: plan the source list, not the schedule. If the publication where the placement will run has its own readers, already covers the topic and would publish the article even without the link, the pace can be whatever the plan requires. If the sources are weak, spreading them out over days does not reduce the risk; it just divides the same risk over a longer period.
Speed alone is not a warning sign. The problem is when speed comes together with other signs: the same anchor, the same template, the same group of interlinked sites and pages with no visitors at all.
Treating an authority score as a decision document
Domain Authority and Domain Rating are not Google metrics. Both are estimates calculated by third parties from the link graph they have collected themselves. Google publishes no authority score that ranks sites.
This distinction directly changes the decision, because a domain can still carry a high score today thanks to links collected years ago, while receiving almost no search traffic. The score measures accumulated history, not current interest.
The right approach: use the score as a pre-screening tool and base the decision on current organic traffic. A site that gets traffic is doing something right on the search side, and that is more direct evidence of health than a third-party estimate. How to build a selection around this criterion is covered step by step in getting backlinks from sites with real traffic.
Where on the site the placement sits also matters. A page in a forgotten archive corner of a high-scoring site and an article in a section that is actually read do not do the same job, even if they appear on the same row in the spreadsheet.
Using the same anchor in every placement
The anchor mistake runs in two directions. At one end are those who put the target commercial keyword on every link; at the other are those who say "keyword links backfire" and turn everything into bare URLs. Both produce the same result: uniformity.
There's no need to guess here; the policy text addresses it directly. Among Google's examples of link spam, links with optimized anchor text in articles distributed on other sites, guest posts and press releases are explicitly listed. In other words, what matters is not the anchor of any single link but the pattern across the whole set.
The right approach: leave the anchor to the language of the sentence. When a real editor mentions a brand, they mostly use the brand name, the page title or a natural phrase describing the topic. The practical check is a single step: if the sentence breaks when the anchor is removed, the anchor is part of the sentence; if the sentence is not affected at all, the anchor was attached afterwards.
Reports that boil down to a single number
Most backlink reports carry a single piece of information: how many placements were made this period. That number is an input, not an output. It shows how much was spent, not what was gained.
This measurement gap has become more expensive now that answers are being generated inside search results. In an Ahrefs study covering about 75,000 brands, the signal with the strongest correlation to brand visibility in AI answers was not the number of links but brand mentions across the web (0.664 versus 0.218). This is a correlation, not causation: it does not prove that increasing mentions leads to appearing in answers. Muck Rack's analysis of millions of citations points in the same direction; most citations come from earned media, while the share of paid and advertorial content stays below 1%.
The right approach: add context measurement alongside the number of placements. In which publication, in which sentence and alongside which topic was the brand mentioned? A link inside a sentence that genuinely describes what you do and a link inside a list of links fill the same row in a report but do not do the same job.
This does not mean paid placements are useless; on the classic search side, a link pointing to the target page still does its job. What it means is this: a plan built solely on paid placements leaves the pool that feeds the answer layer empty. To cover both layers, press releases, industry publications and content with genuine news value should be a separate line item in the plan.
Starting link building before the target page is ready
The most frequently overlooked part of the work is where the link goes. Pointing links at a new page that has not been indexed yet, a thin page of a few sentences or a temporary URL that will be redirected when the campaign ends means wasting the placement.
The reason is simple: a link's contribution to the target page depends on that page being eligible for evaluation. A page that cannot be crawled or has not been indexed is not even a candidate on the search side.
The right approach: stick to the order. The page is published, indexed and starts getting impressions on its own content, and then the link plan kicks in. When this order is kept, the effect of the placement also becomes measurable: if the page is already getting impressions, movement after publication is a readable signal, whereas on a page starting from zero nothing can be isolated.
The cost-free side of the same step is internal linking. Linking to the target page from the site's own related pages before acquiring external links requires no budget, yet most projects skip it and go straight to spending on external links.
Treating a competitor's profile as a prescription
"The competitor has 120 referring domains, we have 60, so let's get 60 more." It sounds like a plan, but it rests on two assumptions, and both are weak.
First, the profile you see on screen is not the competitor's real profile but the part a third-party crawler was able to find. That table does not show which links have been disavowed, which are left over from a site redesign or which come from the brand's news history. Second, the assumption that the competitor owes its current position to that profile is not a tested relationship; it is just two figures sitting side by side.
The right approach: use the competitor's profile as a discovery list, not a list to copy. The useful part is not the number; it is seeing which types of publications genuinely cover the topic and which type of content resonates in them. Targeting the number creates a quota detached from source quality, and a quota always gets filled with the cheapest placements.
Pre-publication decision table
| Common mistake | Evidence-based alternative | Question to ask before publishing |
|---|---|---|
| Spreading placements out by day to look natural | Plan the source list, not the schedule | Would this publication publish this content even without the link? |
| Choosing sites by authority score alone | Use the score for pre-screening and current traffic as the deciding criterion | Does the site get organic traffic today, and on which topics? |
| Using the same commercial anchor in every placement | Leave the anchor to the natural language of the sentence | Does the sentence break when the anchor is removed? |
| Reporting only the number of placements | Also measure the context of the mention | In which sentence does the brand appear? |
| Building links to a page that isn't ready | First get the page indexed and earning impressions | Does the page get impressions on its own? |
| Targeting the competitor's link count | Use the competitor's profile as a discovery list | Does this publication genuinely cover our topic? |
Running these checks one by one before every placement requires the source list, anchor distribution and target page order to be set up in advance. To see how this sequence works in a ready-made plan, look at the package scopes on our backlink packages page.
The four questions we hear most often
How many placements should be made in a period?
There is no absolute number, because the threshold depends on the query, not on your site. What matters is which types of sources mention the pages ranking ahead of you for the target query. The moment the number becomes the target, source selection drops to second place and the decision is handed over to a quota detached from any quality criterion.
When and how do you measure the result of a placement?
Isolating the effect of a single link is often impossible, but that does not mean it cannot be measured. The place to look is not the site's total traffic but the target page's own query set: the number of queries the page receives impressions for and its average position in the weeks after publication. For this reading to be meaningful, the page must also have been getting impressions before publication; otherwise there is no baseline to compare against.
Can placements made with the wrong anchor be fixed?
In most cases, yes, and the cheapest fix is to ask the publisher to change the anchor. What you should not do is delete the target page or change its URL; that move turns the existing placements into either broken links or a lengthening redirect chain. Pause new placements while the fix is under way; otherwise the same pattern keeps growing.
If the budget is tight, which mistake should be avoided first?
The target page mistake. The other mistakes reduce efficiency; this one wastes the budget entirely: nothing measurable remains from a placement made to a page that isn't ready. If the budget is tight, the order is clear: first get the target page ready for publication, then get into a handful of publications that people actually read.
