Branded links vs generic short links: what marketing teams should know

On a Tuesday morning, none of your links work.

You didn't do anything. Somebody you have never met ran a credential-harvesting campaign on the same shortener domain your onboarding emails use, a corporate mail filter noticed, and filters don't block links — they block domains. Your links were never the problem. You just share a postcode with the problem. You'll find out from a customer, and there's nothing to be done except rebuild every link elsewhere and reissue everything you've sent.

That is the actual argument for branded links. It's not about trust, and it is definitely not about click-through rate — which is where these conversations usually start, and where the evidence is thinnest.

What actually differs, and what doesn't

The reputation you're borrowing

A shortener domain is a shared apartment block. Your links live at one address; a few million others live at the rest. Reputation systems — mail filters, safe-browsing lists, carriers, firewalls — operate at the domain level, because that's the only unit they can act on at scale. They cannot evaluate sh0rt.example/9fKq2 on its merits. They decide about sh0rt.example.

So your link's reputation is the average behaviour of every stranger on that domain, and you have no influence over any of it. That isn't bad luck — it's the structural consequence of hiding a destination behind a domain you don't control, which from a filter's point of view is indistinguishable from what an attacker does. It's the same technique.

The deliverability tax

Which leads to the thing nobody warns you about until it's happening: shortened URLs on shared domains are a spam signal in their own right. Put yourself in the filter's position. Hiding where a link goes is a phishing primitive, so a message that hides its destinations behind a domain with mixed history is, on the available evidence, likelier to be hostile than one linking to acme.com. The filter isn't being unfair. It's being correct on the base rates, and you are inside the base rate.

The effect is slow, which is what makes it hard to attribute. Nothing bounces. You just land in more Promotions tabs and quarantines than you otherwise would, and you never see the counterfactual — so it gets diagnosed as subject lines, or send times, or list fatigue, for about a year.

A subdomain of the domain you already send from doesn't carry that signal, because the link now matches the sender. Alignment is the entire point.

You cannot recall a printed QR code

The argument I'd lead with, if I only had one. It has nothing to do with marketing.

# Encoded into 40,000 postcards. Three-year shelf life.
https://go.acme.com/rm26

# The same 40,000 postcards, if someone was in a hurry.
https://sh0rt.example/9fKq2

One of those you can repoint. The other one belongs to somebody else.

A QR code is a URL rendered as ink. Once it's on packaging, a banner, or a piece of direct mail, it is frozen for as long as the object exists — which for packaging is years. You cannot patch it. The only mutable part of that system is DNS, and DNS is only yours if the domain is.

; A branded short domain is one record.
go.acme.com.   300   IN   CNAME   edge.linkprovider.example.

; Which is also the escape hatch: change the right-hand side, and
; every link you ever printed resolves somewhere new, untouched.

Google's own shortener is the cautionary tale. goo.gl stopped accepting new links years ago, and the existing ones were eventually scheduled to stop resolving at all. If Google won't commit to keeping a URL alive indefinitely, nobody will — and every printed asset pointing at a domain you don't own is a bet that a company you have no relationship with stays in a business it may not want.

Where generic short links genuinely win

The honest answer is: more often than a vendor will tell you.

• Throwaway links. A URL pasted into Slack gains precisely nothing from a custom domain.
• You now own the uptime — the trade people forget. A generic shortener has an ops team paid to keep it resolving. On go.acme.com, a lapsed certificate is your outage, on your weekend.
• A brand-new domain has no reputation, which is not the same as a good one. Fresh domains are treated with suspicion by exactly the filters you were trying to please.
• Speed. Thirty seconds versus a ticket with whoever controls the DNS zone, who may not be in your company.

The cost objection is usually the weakest of these — the domain is one you probably already own, and connecting it tends not to be the premium-tier upsell people brace for, so check what your plan actually includes rather than assuming. The honest framing: branded links solve a durability problem. If you don't have one, they solve nothing.

Pick a subdomain, not a clever TLD

If you do go branded, the most consequential decision is one most teams make on aesthetics in about four minutes.

acme.link looks better than go.acme.com. Choose go.acme.com anyway.

A subdomain inherits the history of a domain you've sent mail from for years. A novelty TLD starts at zero, in a neighbourhood with a genuinely poor reputation — cheap TLDs are over-represented in abuse, for the obvious reason that abuse is price-sensitive. Some corporate filters take the shortcut and block entire TLDs.

Which produces the outcome nobody plans for: acme.link can be blocked in more places than the generic shortener you left. You'll have paid for a domain, done the DNS work, and traded a shared reputation for a worse solo one. If it must be a separate domain, use a boring TLD. The branded domain setup is the easy part; the name is the part you cannot undo cheaply once it's printed.

That fifth item is the one people skip, and it's independent of branding: the reason to put a short link in a QR code at all isn't tidiness. It's that the destination stays changeable after the ink dries.

How to test this yourself instead of trusting me

You'll have seen figures claiming branded links lift click-through by some specific and encouraging percentage. Be suspicious of every one of them, including any you see from us. Most come from tests that never controlled for the one variable that decides the question: in most placements, the URL is not visible at all. It's behind a button, an image, an ad card. If nobody can see the link, its domain cannot possibly affect whether they click it. Any lift measured there is measuring the creative.

So branding can only plausibly move click-through where the URL is rendered as text a human reads: plain-text email, SMS, a bio link, a printed asset. A narrow set — but a testable one.

1. Pick a placement where the raw URL is visible. If you can't, stop — there is nothing to test.
2. Split the audience, not the calendar. Branded this week and generic next week measures the weeks.
3. Same destination, same creative, same send time. The only difference is the domain.
4. Keep the UTMs identical except one field — content=branded against content=generic — so both land in the same campaign and you can still tell them apart.
5. Measure clicks from the platform or ESP, conversions from analytics. If those disagree, fix that first: it means the parameters aren't surviving the hop.

That last point matters more than the test. A short link on any domain is a redirect hop, and a redirect hop is invisible to attribution right up until the moment something in the chain drops your query string — at which point the short link gets blamed for a problem it didn't cause. The mechanics of that are worth knowing before you run any experiment, because a test on a chain that eats parameters will produce a confident, clean, entirely fictional result.

The short version: if your links are disposable, use whatever is fastest. If they're going anywhere you cannot recall — print, packaging, a QR code, an email to two hundred thousand people — put them on a domain you control. Not for trust, and not for click-through. Because the alternative is a dependency you have no way to fix, on a morning when you find out from a customer.

And whichever you pick, monitor what the links actually do once they're live. A branded domain resolving to a dead page is just a better-looking version of the same wasted spend.