Ask ten agencies when to move from shared to dedicated cold email infrastructure and eight will answer with a headcount or a revenue number, as if a dedicated IP were a graduation gift for teams that finally got serious. Frank Sondors, who built Salesforge and Mailforge to run sending infrastructure for outbound teams at scale, thinks that framing is backwards.
In a July 21, 2026 post on his Salesforge newsletter, titled "Shared or Private?," Sondors lays out the actual decision: shared infrastructure is where almost every team should start, because it is faster to set up, cheaper to run, and backed by a managed IP pool that someone else is actively protecting. Dedicated infrastructure is not a status upgrade. It is a different risk profile, and it only pays off once specific, checkable conditions are already true.
That distinction matters because most deliverability post-mortems start with the wrong question. A campaign stalls, replies dry up, and the first instinct is to blame the infrastructure — shared IPs are "dirty," a private setup would "protect the sender reputation," and so on. Sondors' argument is that infrastructure is rarely the actual defect. It is just the first thing people point at because it is the most visible knob to turn.
It is also the most expensive knob to turn for no reason. Provisioning dedicated IPs, wiring up new DNS records, and running a full warmup cycle on a fresh domain costs real time and real budget. Spending that budget on infrastructure while the underlying list is still full of dead addresses is a way to guarantee the new setup inherits the exact problem the old one had, just with a longer setup bill attached to it.
Before you blame your cold email infrastructure, run this three-check diagnostic
Sondors is specific about the order of operations. Before anyone touches shared vs. dedicated cold email infrastructure, he says to check three things, because each one produces symptoms that look identical to an infrastructure problem but have nothing to do with which IP the mail left from.
“A dedicated IP does not remove that problem — it just transfers the reputation damage to the sender.”
That line is the center of Sondors' argument. If your list is bad, moving from a shared pool to a dedicated IP does not clean the list. It just means the reputation hit from that bad list now lands on an IP that only you send from, with no other senders' good behavior to dilute it. Shared infrastructure spreads risk across a pool. Dedicated infrastructure concentrates it on you — for better and for worse.
Shared vs. dedicated cold email infrastructure: the actual trade-offs
Once the three-check diagnostic is clean — real data, stable sending pattern, correct technical setup — the shared vs. dedicated decision becomes a straightforward trade-off table, not a values statement about how serious your outbound program is.
| SHARED SENDING INFRASTRUCTURE | DEDICATED IP COLD EMAIL INFRASTRUCTURE | |
|---|---|---|
| Setup cost/time | Low — managed pool, live in hours | Higher — provisioning, DNS, and a full warmup cycle before real volume |
| Control | Limited — you inherit the pool's collective reputation | Full — you own the IP's reputation and every sending decision behind it |
| Risk if used too early | Low — bad behavior gets diluted across other senders in the pool | High — a single bad list or spike tanks a reputation with no dilution |
| Best for | Teams still proving out data quality and sending discipline | Teams with stable volume, tightly managed lists, and active monitoring |
Read the table as a sequence, not a menu. Sondors' recommendation is to start on shared, prove the fundamentals hold, and only then graduate to dedicated — the same volume-driven logic we laid out in the private networks vs. more domains debate, which covers a related but separate question: once you are past the shared-vs-dedicated decision, whether to concentrate sending on fewer private domains or spread it across more of them.
Notice what the "best for" row is not measuring. It is not measuring company size, funding stage, or how long the team has been doing outbound. It is measuring three operational facts: is the data clean, is the volume predictable, and is someone actually watching the health metrics day to day. A five-person startup with a tightly scoped, verified list and disciplined sending cadence is a better candidate for dedicated infrastructure than a hundred-person sales floor running a stale list at inconsistent volume. Size is not the qualifier. Discipline is.
Why dedicated infrastructure concentrates risk instead of removing it
The instinct to reach for dedicated infrastructure early comes from a reasonable place. Shared pools mean sharing an IP's reputation with senders you cannot vet, and one bad actor in the pool can drag everyone's deliverability down with them. That risk is real. But it cuts both ways: a well-run shared pool, actively managed by the provider, also absorbs your own mistakes while you are still learning what "managed" sending discipline looks like.
Dedicated infrastructure removes that safety net. There is no pool to average your reputation against, no other senders' good behavior to offset a spike or a bad list. Everything that hits the IP is attributable to you, and every mailbox provider watching that IP is watching only you. That is exactly what makes dedicated the right call once your fundamentals are solid — full control, no shared blame, and reputation that compounds in your favor over time. It is also exactly what makes it the wrong call while you are still fixing list quality, sending patterns, or DNS authentication, because now every mistake is fully yours and there is nowhere for the damage to dissipate.
Don't diagnose cold email in isolation from other channels
There is a second failure mode Sondors calls out, and it sits next to the infrastructure question rather than inside it: treating cold email deliverability problems as purely an email problem, when the actual gap is that email is being run without any other channel around it. In a follow-up post published July 23, 2026, titled "What I tell people at every conference," he described two experiments he ran that month that make the case directly.
The first: switching LinkedIn follow-up greetings from English-only to localized, native-language openers produced a 29% boost in replies. The second, more pointed experiment: he contacted followers of competitor companies directly on LinkedIn and got roughly a 30% reply rate, with about 20% of those replies positive. The identical audience, contacted cold by email instead, converted at only about 2%.
His takeaway is not "abandon email for LinkedIn." It is that following a competitor already signals category interest — that audience is warmer before you ever send anything, and the channel you reach them through should reflect that. Combining LinkedIn and email against a pre-qualified, competitor-aware audience beat either channel run alone, and it beat cold email by a wide enough margin that the gap can't be explained by infrastructure at all. If a program built for a B2B buying committee is underperforming, the fix is sometimes channel mix, not IP reputation. Before you touch shared vs. dedicated cold email infrastructure, check whether email is even the right first channel for the audience you're targeting.
When to actually move off shared sending infrastructure
Put the diagnostic and the trade-off table together and the decision resolves into a short checklist. Move off shared sending infrastructure when all three are true at once: your list is verified and actively maintained, not a static export from six months ago; your sending volume is stable and predictable rather than spiking after gaps; and your DNS authentication and domain warmup are configured correctly and monitored on an ongoing basis, not set once and forgotten. Miss any one of those and dedicated infrastructure just gives your existing problem a smaller blast radius to concentrate in.
We've seen this play out directly with clients moving from a generic outbound motion to a properly sequenced one — the FMS investor case study is a useful example of what changes when the fundamentals get fixed before the infrastructure does, rather than the other way around. The infrastructure decision came last, not first, and by the time it did it was a formality rather than a gamble.
None of this argues against dedicated infrastructure. It argues against using it as a first move instead of a last one. Sondors' own business exists to run sending infrastructure at scale, and his advice still routes teams toward the cheaper, shared option until they can prove they have earned the more expensive one. That is worth sitting with: the person selling the infrastructure is telling you not to buy the premium version yet. Take the advice at face value.
Do this next: pull your last 90 days of sending data and run the three-check diagnostic yourself before you request a single dedicated IP. Verify list quality against a real bounce and complaint audit, not a gut check. Chart your daily send volume and look for spikes, not averages. Confirm SPF, DKIM, and DMARC are passing and that new mailboxes went through an actual warmup period. If all three come back clean and you're still seeing deliverability problems, dedicated infrastructure is a reasonable next step. If any of them are dirty, fix that first — a dedicated IP will not do it for you.
See where you are cited today
A free snapshot audit of your rankings and AI citations before we ever talk.
Tyler leads work at the intersection of SEO and generative engines at Something Inc., helping B2B brands get ranked and cited across every major AI engine.