TL;DR: There is no dofollow attribute. A link is followed by default and becomes something else when a publisher adds nofollow, sponsored or ugc. Google treats those as hints rather than absolute instructions, which means the clean binary most link building is built on does not exist, and checking what you actually received matters more than arguing about it.
Key Takeaways
Dofollow is not a thing you can request. It is the absence of other attributes.
Four states exist: unmarked, nofollow, sponsored, ugc, each with a different meaning.
The hint change blurred the binary, so followed versus not is no longer a clean split.
Sponsored is the correct marking for anything paid, and asking a publisher to omit it is asking them to misrepresent the relationship.
Anchor text percentages are invented. No published distribution target exists.
Check the markup yourself. View source, find your link, read the attributes.
Anchor targets assume your CMS generates slug IDs from heading text. Confirm on staging and adjust the hrefs if your platform slugifies differently.
There is no dofollow attribute
The brief gets this right and it is worth leading with, because the misconception shapes how people talk to publishers.
The actual markup
An ordinary link is just an anchor element with an href. Nothing marks it as followed, because followed is the default state.
A link becomes something else only when a publisher adds a rel attribute carrying nofollow, sponsored or ugc.
Why the terminology misleads
People ask publishers for a dofollow link, which sounds like requesting a setting. What they are actually asking is for the publisher not to disclose a commercial relationship, or not to apply their site's standard policy.
Phrased that way, the request sounds different, and the second phrasing is the accurate one.
What this means in practice
You want | The accurate request |
|---|---|
An editorial link | Nothing. If they link to you editorially, it is unmarked by default |
A link marked followed | There is no such marking to apply |
A paid link without sponsored | Asking them to misrepresent a paid placement |
Where the confusion comes from
SEO tools report links as dofollow or nofollow, which is a convenient shorthand for present or absent markup. The shorthand then gets mistaken for a thing publishers can toggle on request.
The one case where the request is legitimate
When a publisher applies nofollow site-wide by default, including to genuine editorial references. Asking them to reconsider for an editorial citation is reasonable, and some will.
That is a conversation about their policy rather than a favour.
Why precision matters here
Because the next sections depend on it. Once you see that the states are markings publishers apply rather than settings you request, the rest of the subject becomes considerably clearer.
The four states
What each marking means, as the publisher intends it.
State | Markup | What the publisher is saying |
|---|---|---|
Unmarked | No rel attribute | This is an ordinary editorial reference |
nofollow |
| I do not vouch for this destination |
sponsored |
| This placement was paid for or compensated |
ugc |
| Someone other than the editorial team put this here |
They can combine
A publisher can apply more than one, such as rel="nofollow sponsored" on an advertisement. That is correct usage rather than redundancy.
What each is actually for
nofollow was introduced originally to combat comment spam, and broadened to mean any link the site does not wish to endorse.
sponsored exists so paid relationships can be disclosed to search engines specifically. It is the correct marking for advertising, sponsorship and affiliate links.
ugc covers comments, forum posts and anything contributed by users rather than editors.
The one that matters commercially
Sponsored. If a placement was paid for in money, product or any other consideration, that is the attribute that belongs on it.
Separately from search guidelines, advertising disclosure rules in many jurisdictions apply to the visible content as well, which is a different obligation with its own consequences.
What none of them do
Tell you how much value a link carries. They describe the publisher's relationship to the destination, which is information rather than a quantity.
The practical summary
Four markings, applied by publishers, describing what kind of link this is. Your influence is over what kind of link you earn, not over how it is marked afterwards.
What the hint change did
Google changed its treatment of these attributes from directives to hints, meaning the engine decides how to use the signal rather than obeying it absolutely.
Verify the current wording and link it, since this section's whole argument rests on it.
What it does not mean
Misreading | Correction |
|---|---|
Nofollow links now pass full value | A hint is not an instruction to ignore the marking |
The distinction stopped mattering | Publishers still signal something by applying it |
You can now ignore attributes | They still describe the relationship accurately |
Sponsored links now count | They are explicitly disclosed as paid |
What it does mean
The clean binary that link building was built on does not exist. There is no switch between full value and no value; there is a signal the engine weighs alongside everything else.
Nobody outside Google knows how it is weighed, and anyone claiming a percentage is inventing it.
Why the change happened
Broadly, because blanket nofollowing had made the attribute less informative. Large sites applied it indiscriminately, including to genuine editorial references, so treating it as absolute discarded useful information.
The practical consequence
Stop sorting your link profile into valuable and worthless by attribute. A nofollowed mention in a publication your buyers read is worth more than an unmarked link in a directory nobody visits.
What still holds
An unmarked editorial link from a relevant publication remains the clearest positive signal available. The hint change made the bottom of the scale fuzzier rather than changing the top.
The honest position
Attributes tell you how a publisher characterised the link. They do not tell you its value, and treating them as a value scale was always a simplification that is now visibly wrong.
Which attribute belongs where
From the publisher's side, which is worth understanding because it tells you what to expect and what not to ask for.
The link is | Correct marking |
|---|---|
An editorial reference the writer chose | Unmarked |
Paid placement, sponsorship or affiliate | sponsored |
In a comment or forum post | ugc |
To a site the publisher does not vouch for | nofollow |
Part of a content exchange | sponsored, since consideration was given |
In a press release distributed by wire | Typically nofollow or sponsored |
The exchange case
Content swaps, link exchanges and guest posts offered in return for a link all involve consideration. Whether sponsored applies is a judgement the publisher makes, and the honest answer is usually yes.
This is the case most likely to be argued about and the one where the argument is weakest.
What not to ask for
Removal of a sponsored attribute from a paid placement. That asks the publisher to misrepresent a commercial relationship to search engines, and separately may conflict with advertising disclosure obligations that apply to their own content.
It also tells them what kind of partner you are.
What you can reasonably ask
Ask | Why it is fine |
|---|---|
For a correction to inaccurate information | Always welcome |
For inclusion in a relevant roundup | Normal editorial pitching |
Whether a blanket nofollow policy applies to editorial references | A policy question |
For a specific page to link to | Helpful to them |
If a publisher marks your link sponsored
Accept it. They are describing the arrangement accurately, and the mention still reaches their readers, which was the better reason to want it.
The underlying point
You control what you earn and how you earn it. The marking describes that honestly, and attempting to influence the marking is attempting to change the description rather than the thing.
Anchor text
The text someone uses to link to you. Real, and surrounded by invented precision.
What is genuinely known
Anchor text gives context about the destination, and heavily repeated exact-match anchors across many sites are a recognisable pattern of coordination, because natural linking does not produce that.
That is the whole of what is established.
The percentage matrices
The brief included a distribution target: a specific percentage for branded anchors, another for partial match, another for exact match.
Those numbers are invented. No search engine publishes a target distribution, no study establishes one, and the precision implies knowledge nobody has.
Why they circulate anyway
Because they feel actionable and they give an agency something to report against. A number you can hit is more satisfying than a principle you have to apply.
What actually follows from the mechanism
Principle | Why |
|---|---|
Let publishers write their own anchors | That is what natural looks like |
Do not supply exact-match text to request | That is the pattern being detected |
Expect mostly brand and URL anchors | Which is how people naturally refer to companies |
Do not audit your distribution against a target | There is no target |
The simplest safe approach
Stop specifying anchor text. If a publisher asks what to link, give them the URL and let them write the sentence.
A profile built that way is naturally distributed without anyone managing a distribution, which is the point.
When to actually worry
If you inherited a profile where many sites link with identical commercial phrasing, that is a real pattern worth looking at. It is a cleanup question rather than a ratio to rebalance.
The honest summary
Anchor text matters, the percentages do not exist, and the best way to have a natural profile is not to engineer one.
What PageRank explains
The brief presents the original PageRank formula as current mechanics. Worth separating what the idea explains from what the formula describes.
What the idea gives you
The insight that a link is a kind of endorsement, and that an endorsement from a well-endorsed source counts for more than one from an obscure one. That intuition is sound and still useful.
What the formula is
A description of the algorithm as published in the late 1990s. Modern ranking involves a great many signals, continuous changes, and machine-learned components that no published formula describes.
Quoting the original equation as if it computes present-day rankings is a category error, and it lends false precision to an argument that does not need it.
The part that is still useful
Concept | Still holds |
|---|---|
Links from well-regarded sources count more | Yes, directionally |
A page linking to many things gives less to each | The underlying intuition survives |
Authority is recursive | Conceptually |
The damping factor is 0.85 and this matters to you | No |
Why the last row matters
Nobody is computing their link building strategy from a damping factor. Including it signals technical depth rather than conveying anything actionable, and it is the kind of detail a knowledgeable reader notices as decoration.
On third-party authority metrics
Domain Authority, Domain Rating and similar are vendor estimates built to approximate something Google does not publish. They are useful for rough comparison and they are not the thing itself.
The brief's claim that search engines maintain proprietary equivalents based on the same architecture is a reasonable guess, stated as fact.
What to do with all this
Use the intuition: relevant, well-regarded sources that genuinely chose to mention you are worth more than the alternatives. That is what the mathematics was always describing, and it needs no equation.
Checking what you got
Two minutes per link, and it settles any disagreement about what was applied.
The manual check
Open the page, view the page source, and search for your domain. Read the anchor tag around it: any rel attribute present, and what it contains.
Use view source rather than the browser inspector, since the inspector shows the page after scripts have run and attributes can be modified client-side.
What to record
Field | Why |
|---|---|
The URL of the linking page | Not just the domain |
The attributes present | The actual markup |
The anchor text used | What they chose to write |
The sentence around it | Whether it describes you usefully |
The date checked | Links change |
The fourth row matters more than the attributes. A mention describing what you do and who you suit does work that a bare link does not, which the outreach article covers.
Links change
Pages get rewritten, sites get redesigned, attributes get added during a policy change. A link checked once is a link checked once.
Quarterly is a reasonable cadence for anything you worked to earn.
When a link disappears
Frequently recoverable with a polite note, especially if the page was rewritten rather than the link deliberately removed. That note is cheaper than earning a replacement.
When an attribute changes
Usually a site-wide policy change rather than anything about you. Worth noting, not worth a complaint, and the mention still reaches their readers.
What not to do with the data
Sort your profile into valuable and worthless by attribute. Since the markings are hints and tell you about the publisher's relationship rather than the link's value, that sort produces a misleading picture.
Sort by whether the source matters to your buyers instead.
Six mistakes
1. Asking a publisher for a dofollow link. There is no such attribute. What you are asking is for them not to apply their policy, or not to disclose a paid relationship.
2. Asking for sponsored to be removed. That asks them to misrepresent a commercial arrangement, and it tells them exactly what kind of partner you are.
3. Reading the hint change as nofollow now passing value. It means the binary is fuzzy, not that the marking is meaningless. Nobody outside Google knows the weighting.
4. Chasing an anchor text distribution. The percentage matrices circulating are invented. Let publishers write their own anchors and the profile is natural without management.
5. Quoting the original PageRank formula. It describes an algorithm from the late 1990s. The intuition survives; the equation does not compute present rankings.
6. Sorting your profile by attribute. A nofollowed mention in a publication your buyers read beats an unmarked link in a directory nobody visits.
The pattern
Most come from treating attributes as a value scale rather than as a description of the publisher's relationship to the link.
The one to do this week
Pick five links you worked to earn, open each page, view source, and read the actual markup and the sentence around your link.
The sentence will tell you more than the attribute does, and if it does not describe what you do, that is the thing worth fixing.
FAQs
Is there a dofollow attribute in HTML?
No. Links are followed by default, and a link becomes something else only when a publisher adds nofollow, sponsored or ugc. Dofollow is shorthand for the absence of those markings.
What do the four link states mean?
Unmarked means an ordinary editorial reference. Nofollow means the publisher does not vouch for the destination. Sponsored means the placement was paid for. Ugc means a user rather than an editor added it.
Do nofollow links pass value now?
Google treats the attribute as a hint rather than an absolute instruction, which makes the old binary fuzzy. Nobody outside Google knows how it is weighted, and any specific percentage is invented.
Can I ask a publisher to remove a sponsored attribute?
You can, and you should not. It asks them to misrepresent a commercial relationship, and advertising disclosure obligations may apply to their content separately from search guidelines.
What anchor text distribution should I aim for?
None. The percentage matrices in circulation are invented and no search engine publishes a target. Let publishers write their own anchors and the distribution takes care of itself.
Is the PageRank formula still how ranking works?
The intuition holds: endorsements from well-regarded sources count more. The published equation describes an algorithm from the late 1990s and does not compute present-day rankings.
Are Domain Authority and Domain Rating official?
No. They are vendor estimates approximating something search engines do not publish. Useful for rough comparison, not the thing itself.
How do I check what attribute my link has?
View the page source, find your domain, and read the anchor tag. Use view source rather than the inspector, since attributes can be modified after the page loads.
Read five of your own links
Pick five you worked to earn. Open each page, view source, find your domain, and read the anchor tag and the sentence around it.
The sentence will tell you more than the attribute does. If it does not describe what you do and who you suit, that is the thing worth fixing, and most publishers will accept a correction.
The working rules
Stop asking for dofollow. There is no such attribute to apply.
Accept sponsored on paid placements. It describes the arrangement accurately.
Let publishers write the anchor. That is what a natural profile looks like.
Judge links by the source and the sentence, not by the markup.
Re-check quarterly. Pages get rewritten and attributes change.
Ask for corrections when a source describes you wrongly. Fastest lever available.
When the constraint is having something worth citing
It usually is. Blazly SEO covers research, planning, drafting and direct publishing from about $39 a month.
Summary
There is no dofollow attribute: links are followed by default and become something else only when a publisher adds nofollow, sponsored or ugc, each describing their relationship to the destination rather than setting a value. Google treats those markings as hints rather than absolute instructions, which means the clean binary most link building was built on does not exist, and nobody outside Google knows the weighting. Sponsored is the correct marking for any paid placement, and asking for its removal asks a publisher to misrepresent a commercial relationship. Anchor text matters and the percentage distributions in circulation are invented, so the safest approach is to let publishers write their own. Judge a link by the source and by the sentence around it, both of which you can read yourself in two minutes.