SuperSend

Two narrow model features on a product that is otherwise infrastructure engineering.

The first labels replies automatically in the unified inbox so interested buyers surface above the noise. The second categorises bounce failures by cause, separating a bad address from a server block from a temporary failure, which is genuinely useful because those three demand different responses and most platforms report them as one number.

That is the whole of it. Two passes across the pricing page, the feature navigation and independent coverage located no generated copy, no personalisation engine, no research agent, no lead scoring and no assistant.

Everything the customer is actually buying is deterministic: dedicated servers and address pools, sender provisioning, authentication record configuration, warm up phases, rotation, throttling, adaptive pacing, placement testing and sequencing logic. None of it requires inference and all of it requires careful engineering.

That is a coherent position rather than a gap. A platform whose promise is that mail arrives is better served by predictable behaviour than by clever behaviour, and a customer can reason about what it will do.

The bounce categorisation is the one place where inference earns its keep, because failure reasons arrive as unstructured server responses.

Ask whether any generated content capability is planned.

The system restrains itself on evidence from the receiving end, which is the right trigger.

Delivery pacing adapts to live acceptance and deferral signals. When a receiving server defers or throttles, the platform slows against that destination rather than continuing on a fixed schedule and accumulating damage. That is the mail transfer protocol equivalent of listening rather than shouting, and almost nothing in this category describes doing it.

Routing is equally adaptive: sends are directed through the healthiest inboxes in real time rather than round robin, so a struggling sender is bypassed automatically. Per domain tuning applies different treatment to different destinations. Bounce analysis categorises failures so the correct remedy follows.

An acceptable use policy and an abuse reporting route are published as separate documents, which together are the governance layer most platforms in this category omit entirely.

On the campaign side, conditional routing lets sequences branch on behaviour, and interface calls can fire mid sequence to update external systems.

What is unspecified: no audit trail is described, no permission model governs who may launch campaigns, and nothing states what happens when an acceptable use breach is detected.

Ask what triggers an acceptable use enforcement action and what follows.

Two model features and no technology named behind either.

Reply labelling reads inbound correspondence and assigns a category so a seller knows which messages matter. Bounce analysis interprets server responses and classifies the cause of failure. Two passes located no model provider, no model name, no version, no statement of what reply content is transmitted for inference, no training or retention position and no accuracy figure for either.

The reply labelling is where the omission has operational weight, because a mislabelled positive reply sits unread in the wrong bucket and the seller never knows a live opportunity was missed. No error rate is published.

The surface is genuinely small, so this matters less than it would on a platform whose agents write and send. Nothing model driven here reaches a prospect: both features operate on material already returned to the customer.

The contrast with the rest of this vendor's disclosure is the point worth recording. It publishes a subprocessor list, a retention policy, a security measures document, a processing agreement and an acceptable use policy. On the two model features it publishes nothing, and a subprocessor list is exactly where the inference provider would ordinarily appear.

Ask whether the reply labelling provider appears on the published subprocessor list.

One named testimonial with a figure, and a public admission about churn that no other vendor here would make.

The testimonial is specific: a named user reports four to five introductory meetings a week from cold outreach for what he describes as low effort. A product launch platform carries substantive user commentary praising rotating senders, warming campaigns, native validation and retargeting.

The unusual evidence is a post on the company's own professional network page. In it the vendor states that the product naturally has high churn, that it is not perfect, that it is a complex business critical product with a great deal of competition and little tolerance for inefficiency, and that its average customer lifetime was three and a half months of recurring revenue before a pricing experiment. Publishing your own churn profile and average customer lifetime is not evidence of outcomes, but it is a level of candour that makes the rest of the vendor's claims more credible rather than less.

What two passes could not locate: named customer logos, case studies, a review count or rating, and any measured deliverability or reply rate figure. The vendor states it sends millions of messages monthly for customers with no basis given.

Ask for placement rate data across the customer base.

A postmaster page and an abuse reporting route, which are what infrastructure operators publish and cold email tools do not.

Those two are the finding. A postmaster page is the address mailbox providers and network operators use to raise delivery and abuse concerns with a sending operator, and publishing one is an acknowledgement that the vendor is part of the mail ecosystem with obligations to it rather than merely a customer of it. An abuse reporting route gives recipients and third parties a way to report misuse that reaches the party who can actually stop it, since this vendor operates the servers and the addresses.

Across this entire index no other vendor publishes either.

Around them sits the fullest governance set recorded here: an acceptable use policy defining what customers may not do, a compliance hub, a processing agreement, a subprocessor list, a data retention policy and a security measures document, each published separately rather than folded into one page.

The architecture reinforces it. Because the vendor owns the servers and addresses rather than renting shared pools, it can enforce its own policy directly and bears the reputational consequence when a customer abuses the platform.

What is absent: no unsubscribe mechanism is named in the feature set, and no consent basis is stated for the lists customers bring.

Ask what the acceptable use policy prohibits and how breaches are enforced.

Four separate governance documents published, covering exactly what a buyer's counsel asks for.

A data processing agreement, a subprocessor list, a data retention policy and a security measures document, each as its own linked page rather than a paragraph inside a privacy policy. A compliance hub sits alongside them, with privacy policy and terms beneath.

The subprocessor list is the one that separates this record. Across this index most vendors publish nothing about who else touches customer data, several claim certifications without naming a scope, and a handful disclose suppliers accidentally through pricing tables. Publishing the list as a maintained document is the transparent form, and for an infrastructure operator running dedicated servers and address pools it is the document that reveals the actual supply chain.

A retention policy published separately is nearly as rare, and it is the question every buyer eventually asks about reply correspondence and campaign history.

None of the four was opened in this pass, so their contents are unrecorded and a buyer should read them.

What is held is substantial: customer contact lists, reply correspondence across three channels, sending telemetry per domain, and provisioned domains and mailboxes held on the customer's behalf.

Ask which subprocessors are listed and what the retention period is for reply data.

The vendor supplies no contact data, and the credits meter verification rather than supply.

Two passes across the pricing page, the feature navigation and independent coverage located no contact database, no record count, no lead finder, no enrichment offering and no third party data supplier anywhere in the product. Customers bring their own lists or connect their own sources.

The credit system reinforces the position rather than obscuring it. Credits are consumed by address validation, at one per address, and by placement testing, at five per seed. Both are quality operations performed on records the customer already holds, not lookups against a pool the vendor assembled. A buyer is paying to check their own data rather than to rent someone else's.

That is clean by construction, and it removes the consent question that dominates this axis elsewhere in the index. Whoever obtained the list carries the obligation, which is where it belongs.

The one unexplained element is the validation method itself. Verifying an address at scale requires infrastructure and, usually, upstream services, and nothing describes how it is performed or what accuracy it achieves. The published subprocessor list would be the place to look.

Ask how address validation is performed and at what measured accuracy.

The vendor describes its own professional network automation more honestly than anyone else in this index.

In its own comparison content it states that network steps run as browser based automation, and adds parenthetically that this is not a magic interface bypassing how the platform actually works. That is a vendor declining to dress up what it does. Set against another record built this session marketing randomised limits that make automation undetectable, the difference in posture is considerable, and it deserves recording even though the underlying activity is the same.

What that activity is: connection requests, profile visits, messages and follow ups, sold as an addition at sixty nine dollars per connected sender, alongside outreach on a microblogging platform. Both are prohibited by those platforms' terms whoever performs them and however candidly described.

The email side carries no exposure at all. The vendor operates its own servers and address pools, provisions its own sender identities and configures authentication records directly, which is entirely sanctioned infrastructure.

No scraping, no account renting and no evasion marketing appears anywhere.

The honest summary is real exposure on two channels, described without pretence, alongside a fully owned email path.

Ask what happens to a customer's sequences when a connected network account is restricted.

The stewardship documentation is the deepest in this index and no independent attestation stands behind it.

Published separately: a security measures document, a processing agreement, a subprocessor list, a data retention policy, a compliance hub, an acceptable use policy, an abuse reporting route and a postmaster page, with an operational status page alongside. Eight governance documents is more than any other record here, and the set covers the questions a buyer's security and legal functions actually ask.

The model half carries almost no exposure. Reply labelling and bounce categorisation both operate on material already returned to the customer, so nothing generated reaches a prospect and a failure produces a misfiled message rather than an embarrassing send. No provider, evaluation or accuracy figure is published for either.

What holds this below the top band is verification. Two passes located no certification of any kind, no audit report, no penetration test and no vulnerability disclosure route. The abuse and postmaster routes handle mail problems rather than security research findings, which is a different function.

For an operator running dedicated servers and address pools on customers' behalf, an independent examination is the natural next step.

Ask whether any independent security assessment has been completed.

A recipient has somewhere to complain that reaches the party who can act, which is rare here.

The abuse reporting route and postmaster page mean someone receiving unwanted mail from a customer of this platform can raise it with the operator of the sending infrastructure rather than with the sender who is ignoring them. Because this vendor owns the servers and addresses and publishes an acceptable use policy, that complaint has somewhere to land and something behind it. Across this index that combination exists nowhere else.

Identity is genuine throughout. Mail leaves from the customer's own domains, or from domains and mailboxes purchased in their name, with authentication records configured correctly. Two passes located no persona, no rented account, no synthetic sender and no undetectability marketing.

Two gaps keep it mid band. Profile visits on the professional network generate a notification telling the recipient a named person looked at their profile when software did, which is the same manufactured signal recorded elsewhere this session. And no unsubscribe mechanism appears anywhere in the feature set, on a platform sending at volume.

Reply labelling operates on inbound mail and touches nothing the recipient sees.

Ask what unsubscribe handling is built into sequences.

Interface documentation on its own subdomain, a protocol server with its own page, and interface calls that fire inside a sequence.

That last one is the distinguishing capability. Sequences can include steps that call an external system mid flow to sync deals, tags and activity in real time, which means the platform is not only reachable from outside but reaches outward at defined points in a campaign. A sequence becomes a workflow rather than a send schedule.

The interface itself is documented openly on a dedicated subdomain rather than gated behind a sales conversation. A model context protocol server has its own resource page, making this the ninth such server recorded in this sweep.

Native integrations cover two named record systems, with two general automation platforms and webhooks alongside for everything else.

An operational status page reports live, which for an infrastructure operator is not optional and is nonetheless absent from most records here.

The estate is completed by a blog, a careers page and a public review invitation.

What could not be established: interface rate limits, authentication scheme and what the protocol server exposes, since neither the documentation nor the protocol page was opened.

Ask what the protocol server exposes and what the interface rate limits are.

Dedicated servers and isolated address pools, stated as the architecture rather than as an upgrade.

The vendor describes operating its own sending infrastructure with dedicated servers, isolated address pools, sender identity provisioning and redirect domains, scoped to each customer's volume. Isolation is the substantive claim: reputation is not shared with other customers, so one account's behaviour cannot damage another's placement. That is the same architectural position taken by one other record this session and it is the correct one for this category.

The managed track goes further, with the vendor operating the infrastructure directly and assigning a technical account manager.

An operational status page reports live, and a security measures document is published separately.

What two passes could not locate: a hosting provider, a region, a data centre, a residency commitment, a backup position or a continuity plan. For a European buyer that is the unresolved question, since dedicated infrastructure says nothing about where it physically sits, and the platform holds reply correspondence and contact lists as well as sending telemetry.

The published subprocessor list would be the natural place to find the underlying providers and was not opened.

Ask in which regions dedicated infrastructure can be provisioned.

A trust centre exists in substance, distributed across eight documents rather than gathered on one page.

What is published: a security measures document, a compliance hub, a processing agreement, a subprocessor list, a data retention policy, an acceptable use policy, an abuse reporting route and a postmaster page, with an operational status page alongside. A buyer running a supplier review would find answers to most of their questions, which is more than can be said for the majority of records in this index.

The compliance hub in particular suggests the vendor expects to be asked and has organised for it.

What is missing is independent verification. Two passes located no certification of any kind, no audit report, no examination, no penetration test result, no auditor name and no vulnerability disclosure route through which a researcher could report a flaw. Every assurance is the vendor's own statement, however thorough.

That gap matters here more than it would elsewhere, because this vendor operates sending infrastructure on customers' behalf and holds provisioned domains and mailboxes in their name. Compromise would affect the customer's own sending identity rather than only their data.

The documents were not opened in this pass and may name a certification.

Ask whether an independent examination exists and where a security researcher should report findings.

This vendor published unusually good pricing and has just removed it.

The current page carries no figures. It states that pricing is scoped to volume, comprising a flat platform fee plus a per send rate, and that both are scoped on the first call. The only action available is a pricing request.

What was there until recently is recoverable and worth recording, because it was better than most of this category manages. Two named tiers with per message rates published to four decimal places, at four tenths of a cent and a quarter of a cent per send. An entry point at ninety nine dollars monthly. Network senders at sixty nine dollars each. Unlimited contacts on every plan. Credits explained precisely, at one per address validation and five per placement test seed. Usage above the plan billed at month end with no add on blocks to purchase.

And one line no other vendor in this index has published: an estimate of what customers typically spend in total, around three hundred and fifty seven dollars monthly once inboxes and domains are added. A vendor volunteering its own realistic all in figure, rather than the headline, is the most useful thing a buyer can be told.

All of it is now behind a call. This is the second vendor this session to withdraw published pricing.

Ask for the platform fee and per send rate in writing before the call.

A published retention policy, which is the exit document almost nobody writes.

Most vendors in this index describe how to get data in and say nothing about what happens to it afterwards. This one publishes a data retention policy as a separate linked document, which is where a buyer finds out what is kept, for how long and what is deleted on departure. It was not opened in this pass and a buyer should read it, but its existence is the point.

Three programmatic routes out exist alongside it: a documented interface on its own subdomain, webhooks, and a protocol server. Sequences can also call external systems mid flow, so campaign outcomes reach a customer's own record system as the work happens rather than only at export.

The infrastructure position is mixed. A customer who brought their own domains and mailboxes keeps them outright. A customer who bought them through the platform holds assets provisioned in their name, and whether administrative control transfers on departure is unstated. Dedicated address reputation, built over months of warm up and sending, belongs to infrastructure the vendor operates and does not travel.

No export scope statement or account closure process was located.

Ask whether purchased domains and mailboxes transfer on cancellation.

Pacing that responds to what receiving servers actually say is the most sophisticated mechanism recorded on this axis.

Most platforms send on a schedule and discover problems from bounce reports afterwards. This one adapts pacing from live acceptance and deferral signals, meaning it reads the responses receiving mail servers return during a send and slows against that destination when it is being pushed back. Deferrals are the earliest available warning that a receiver is unhappy, arriving well before a block, and reacting to them is the difference between backing off and burning a domain. Alongside it, tuning is applied per receiving domain rather than uniformly, which is the correct granularity since each major provider behaves differently.

The rest is comprehensive. Dedicated servers and isolated address pools so reputation is never shared. Automatic configuration of sender policy, key signing and reporting records on provisioned domains. Warm up on every plan in two phases, an initial ramp and continuous background warming so mailboxes do not decay when campaigns pause. Domain, subdomain and inbox rotation routing each send through the healthiest available sender in real time. Throttling. Placement testing against four named providers. Bounce analysis separating bad addresses from server blocks from temporary failures. Sender health monitoring.

A postmaster page completes the position.

Ask what the per domain pacing responds to and how quickly it adjusts.

Two named verticals, both regulated financial services, and a gap in the middle of the range.

The verticals are investment banking and business lending, each with its own page. That is an unusual pair for a cold email platform and a revealing one: both are relationship led, both are heavily regulated, and both involve outreach where deliverability failure is expensive because the addressable universe is small. A platform choosing those two is selling precision rather than volume.

The volume banding is where the coverage problem sits. The self serve track serves teams below a million messages monthly. The managed track begins at a million. Independent review notes the gap between the two is wide with no middle ground, and that a team sending eight hundred thousand messages monthly is still on self serve, managing their own warm up, reputation monitoring and placement testing at a volume where that is substantial operational work.

That is a real segmentation weakness rather than a marketing one, and it lands on exactly the customers growing fastest.

What is absent: no geography, no company size band, and no statement of which markets the infrastructure serves.

Ask what support exists for teams between two hundred thousand and a million messages monthly.