How many links should a newsletter have?
TL;DR
Link count is not a deliverability variable. Google's sender requirements never mention it, and the blocklists that matter check the reputation of your link domains, not how many anchors you shipped. The real cost of extra links is attention, and the number to watch is words per editorial link, after you subtract the header, footer and social boilerplate that inflates every automated count.
Ask how many links should a newsletter have and you get two answers that cannot both be right. The conversion crowd says two to five, because every extra link splits attention. The deliverability crowd says keep it low or you will trip a filter. We parse a lot of other people's newsletters here, and the second answer is folklore that outlived the filters that inspired it. The first is closer, but it counts the wrong things.
The version we would defend: link count tells you almost nothing on its own, because most of the links in a newsletter were never an editorial decision. Strip those out and the number that remains is worth arguing about. It still has nothing to do with spam.
How many links should a newsletter have, by type
No universal ceiling exists, so here are the ranges we work to. A curated editorial issue, the Morning Brew or daily-digest shape, carries 20 to 40 editorial links comfortably. That is the format working as intended. A DTC promotion wants three to eight distinct destinations, though the same one often appears five times behind an image, a headline and a button. B2B lifecycle email sits at one or two. A creator newsletter with an essay and sign-off lands near five to fifteen.
These ranges sit badly against the advice on page one of Google. "Keep it under five links" is decent guidance for a lifecycle email and actively harmful for a digest, where the links are the product. Anyone giving one number across all newsletter types has not read enough newsletters.
How many links your newsletter really has
Most of the links in a newsletter were never an editorial choice. This is the part nobody writes about, and it changes the question.
When our extractor processes an issue, it counts every anchor carrying an href in the HTML body. That is what your spam checker does too, and your ESP's preflight, and every "too many links" warning you have had. It cannot tell a link you agonised over from the Instagram glyph in your footer.
Add up what a normal template contributes before you write a word: logo, view in browser, three or four nav items, four to six social icons, unsubscribe, preference centre, privacy policy, address block, app store badges. That is fifteen to twenty links of furniture. A send reporting 53 links might hold 30 editorial ones, and 30 is the only figure worth reasoning about.
We spot-checked the retail sends sitting in our ingestion queue while writing this. Not a benchmark, just a handful of real issues from two footwear brands. Raw counts ran 3 to 53 anchors per issue against 68 to 666 words of copy, working out at 12.5 words per link at the densest and 25 at the loosest. Every figure there is inflated by template chrome. That is the point: take an automated link count at face value and you will trim real content to compensate for a footer you are legally required to keep.
Link count does not trigger spam filters, link reputation does
Google's sender guidelines are public and specific. Authenticate with SPF and DKIM, publish DMARC, keep valid PTR records, use TLS, hold spam complaints below 0.30%, comply with RFC 5322, support one-click unsubscribe. Link count is not on the list. Neither is link-to-text ratio. The one piece of link guidance is that links should be visible and easy to understand, so recipients know what to expect when they click.
That last clause is the real tell. Filters care about deception, not arithmetic. A link whose visible text says one thing while the href points somewhere unrelated is a classic phishing signature, and it will do more damage than thirty honest links ever will.
What gets evaluated is domain reputation. Spamhaus DBL and URIBL are domain blocklists: they pull the domains out of a message body and check them against a list. The arithmetic runs opposite to the folklore. Forty links to your own clean domain is forty lookups that come back clean. One affiliate redirect chain terminating somewhere you have never checked is a single lookup that comes back bad. Count is not the variable. Membership is.
So a 40-link digest lands in Gmail's Primary tab while a six-link promotion goes to spam. The digest links to reputable publishers from a warm domain with high engagement. The promotion routes through two redirect hops and a shortener. The rest of that machinery is in our guide to how spam filters score a newsletter, and our spam score checker flags redirect chains and mismatched anchor text in an issue you paste in.
The link-to-text ratio that does matter
Drop the spam framing and the ratio becomes useful again, because it measures reading experience. The rule doing the rounds is one link per 125 words. It came from prose-heavy email and holds up reasonably there. The arithmetic: divide body word count by editorial links. A 900-word essay with seven links runs 129 words per link and reads fine. The same 900 words with 36 links runs 25 words per link, a link every sentence and a half. At that density readers stop treating any of them as a recommendation.
But applying 125 to a curation newsletter is nonsense. A digest built on twelve links to twelve stories runs 30 to 60 words per link by design, and clicks fine. The ratio is a diagnostic for prose, not a rule for lists. Our working ceiling for anything with a conversion goal is 100 words per editorial link. Below 25, accept that you have built a link list and design it as one.
Link density compounds with image density: a heavily imaged send has less text to carry its links, so the ratio degrades from both ends. The visual half of that is in the newsletter image-to-text ratio guide.
What extra links cost you is the primary CTA
Here is the real trade. Every extra destination competes with the one you care about. Not for space. For the reader's single decision.
Repetition and competition get conflated constantly. Three links to the same destination is repetition, and it lifts clicks there because you caught the reader at three scroll depths. Three links to three destinations is competition, and it splits them. A hero button, a mid-body text link and a P.S. all pointing at one page beats a hero button, a blog link and a webinar link, even though both send three links.
One technical detail we keep hitting, which says something about how email is built: our extractor counts <button> elements, and across real newsletters that count is almost always zero. Nobody uses button elements in email. Every button you have clicked in an inbox is a table cell with a background colour or a padded anchor, because Outlook's rendering engine left designers no reliable alternative. Any tool that counts your CTAs by looking for buttons is counting nothing.
Editorial newsletters get an exemption here, and they should. A digest has no single conversion goal, so twelve equal-weight links is the format rather than a focus failure. The mistake is running an editorial layout and expecting a promotional click rate from one buried CTA. Pick a shape. Our click-to-open rate benchmarks by niche break the numbers down by send type, and the CTA friction guide covers what happens after the click.
See whether your links are competing with each other
Our CTA analyser reads the anchors in an issue, groups them by destination, and shows how many separate actions you are really asking for.
Analyse your CTAs →Your ESP rewrites the links before anyone sees them
One more layer sits between your draft and the filter. Mailchimp, Klaviyo, Beehiiv, Substack and ConvertKit all rewrite outbound hrefs through their own tracking infrastructure when click tracking is on, the default nearly everywhere.
The consequence is neat. Your 40 links to 40 publishers do not reach the filter as 40 domains. They arrive as 40 URLs on one or two tracking domains belonging to your ESP, which collapses the domain-diversity signal and moves the reputation question onto your ESP's click domain rather than the sites you linked to.
That cuts both ways. A shared tracking domain inherits the behaviour of everyone else on it. Dedicated click domains cost more and need warming, but the reputation is yours. For most senders under a few hundred thousand a month, shared is still the right call.
Rewriting has an intelligence use too. In a competitor's issue, the tracking domain names their ESP before you read a header, and the query strings often expose internal campaign naming. We cover that in reverse-engineering a competitor newsletter, with the tagging half in the UTM parameters guide. For doing teardowns at scale, our MailCharts comparison covers the options.
Audit your own link density in ten minutes
Open your last send, view source, and count occurrences of href= in the body. That is your raw number, and it will be higher than you expect. Subtract the furniture: logo, view in browser, nav, social icons, unsubscribe, preferences, privacy, address, app badges. What remains is your editorial link count. Divide body word count by it, and compare against 100 words per link if the send has a conversion goal.
Then group them by destination. If more than one carries real weight, you have more than one CTA, whatever the brief said.
When you cut, the order is reliable. Duplicate destinations that add no new scroll position go first. Social icons next, because they route subscribers to a platform that will not deliver your next issue either. Then redundant "read more" links sitting beside a headline that already links to the same article. Then anything pointing at a domain you have not checked this quarter, usually a stale affiliate link or a partner who let their domain lapse.
What survives that pass is the newsletter. The number it adds up to matters far less than whether every link in it earned its place.
Frequently asked questions
How many links should a newsletter have?
It depends on the format, and the honest ranges are wider than most advice admits. A curated editorial issue carries 20 to 40 editorial links without trouble. A retail or DTC promotion works best with three to eight distinct destinations, repeated across image and text. A B2B lifecycle email should have one or two. Count only the links you chose to put in the body, because logo, social icons, unsubscribe and preference links inflate every automated tally without changing how the issue reads.
Do more links in an email increase spam risk?
Not by count alone. Google's email sender guidelines list authentication, a spam complaint rate below 0.30%, valid reverse DNS, TLS and one-click unsubscribe, and never mention link count or link-to-text ratio. Their only link guidance is that links be visible and easy to understand. What filters evaluate is domain reputation: blocklists such as Spamhaus DBL and URIBL list the domains found in message bodies, so one link to a burned domain hurts more than forty links to clean ones.
What is a good link-to-text ratio for email?
The repeated rule is one link per 125 words, a reasonable ceiling for prose-heavy newsletters but wrong for curation formats where each link is the content. Better method: divide body word count by editorial links, after subtracting header, footer and social boilerplate. Above roughly 100 words per link you have room to add. Below about 25 you are writing a link list, which is fine for a digest and poor for anything with a single conversion goal.
Does the unsubscribe link count toward your link total?
It counts in any automated tally, including ours, because parsers count every anchor with an href in the body. It should not count in your editorial decisions. Unsubscribe, preference centre, physical address, view-in-browser and social icons are required or near-required furniture. Strip them out before judging whether an issue has too many links, or you will cut real content to offset boilerplate you cannot remove anyway.
How many CTAs should one newsletter have?
One primary action, repeated up to three times in different formats, beats three competing actions. Repeating a destination raises total clicks on it. Competing destinations split them. Editorial newsletters are the exception: a curation format has no single conversion goal, so twelve equal-weight links is the product rather than a failure of focus. The mistake is running an editorial layout while expecting a promotional click rate from one buried button.