Telegram Bulk Message Sender: How to Pick One That Works

Search for a Telegram bulk message sender and you get a list of things that look like competitors. They are not. Hidden inside that one phrase are five different machines built on two incompatible halves of Telegram, and the half a tool sits on decides something no feature list will tell you: whether it can reach the people you actually want to reach. Pick from the wrong half and nothing else matters. The rotation is irrelevant, the message spinning is irrelevant, the dashboard is beautiful and the messages go nowhere.

This guide is about the choice, not the campaign. If you already know which tool you are using and you want the step by step of running one, the mass DM walkthrough covers that end. Here we stay on the question people ask first and answer last: which sender, and how do I tell before I pay.

Every platform limit below comes from Telegram's own documentation and is quoted as Telegram states it. Where a number is ours, measured on our own accounts, it says so. Where nobody publishes a number, this guide says that instead of inventing one.

What a bulk message sender actually is

Strip the marketing away and a bulk sender does four things: it holds a list of recipients, it holds one or more identities to send from, it decides the order and the spacing, and it writes down what happened. Everything else on the feature page is a variation on those four.

The category feels confusing because the second item, the identity, can be two completely different things. A tool can send as a bot, or it can send as a user account. These are not two settings on one road. They are two roads, built by Telegram for two different purposes, with different rules about who you may contact, how fast, and what happens when someone objects.

The four moving parts

The list. Usernames, phone numbers, numeric IDs, group members, or subscribers who opted in. Where the list came from matters more than most people expect, and it gets its own section further down.

The identity. A bot token, or one or more logged in accounts. This is the fork, and almost every disappointment in this category traces back to it.

The pacing. Delay between messages, the order of sending, how many identities share the work, and what happens after an error.

The record. Sent, failed, blocked, replied. A sender that cannot tell you which of those happened to each recipient is not a sender, it is a script with a window.

Why the same words describe very different products

A browser extension that clicks buttons inside Telegram Web is a bulk sender. So is a hosted platform running dozens of accounts on separate connections. So is a fifteen line script somebody published on a code sharing site. All three appear under the same search phrase, and the gap between them is not quality, it is category. Comparing them on price is like comparing a bicycle to a bus ticket.

The one question that decides everything: bot or account

Before comparing anything else, answer this: can the people on your list be reached by a bot at all? For most outreach lists the answer is no, and that single fact removes half the tools on the market from consideration.

What a bot can reach

A bot can message people who already have a chat with it. That is the design. Someone taps start, or joins a group where the bot lives, and from then on the bot may write to them. A bot cannot pick a username off the internet and open a conversation. This is not a rate limit that a better tool or a bigger budget can raise. It is the shape of the Bot API.

So if your list is subscribers, customers, people who signed up through your bot, support contacts, anyone who has already pressed a button of yours, the bot road is open, and it is by far the calmest road available.

If your list is group members you collected, usernames from someone else's channel, or people who have never heard of you, the bot road is closed before you start. No configuration opens it.

What a user account can reach

A logged in account can message a stranger the same way a person can: find the username, open the chat, write. The client libraries built on Telegram's user protocol expose what the app exposes, which is why every cold outreach tool on the market lives on that side.

That reach comes with the rest of being a person on Telegram, and the rest is the second half of this guide: reports, limits, human moderators, and a ceiling that no tool sets and no tool can raise.

How to tell which side a tool is on before you pay

Ask one question: does it want a bot token, or a phone number and a login code? A token means bot. A phone number and a code means account. If the answer is a phone number and the seller also promises you can reach anyone on Telegram, the promise is at least architecturally possible. If the answer is a token and the seller promises the same thing, the seller is either confused or lying, and you learned that for the price of one question.

Route one: broadcasting with the Bot API

This is the road Telegram built for what many companies actually need, and it is chronically underused because the phrase people search for is bulk sender, not broadcast.

The published limits

Telegram states the bot broadcasting limits plainly in its own bot documentation. In a single chat, avoid more than one message per second. In a group, a bot cannot send more than twenty messages per minute. For bulk notifications, a bot cannot broadcast more than about thirty messages per second.

Thirty per second is a bigger number than it looks. That is one hundred and eight thousand messages an hour, to people who asked for them, with no accounts to warm up, no connections to buy, and nothing that can be restricted. For a subscriber list the bot road is not the compromise. It is the good road.

Paid broadcasts, and the gate in front of them

Telegram also sells a higher ceiling. Enabling paid broadcasts through the bot management bot raises the limit to a thousand messages per second, and each message above the free thirty per second costs a tenth of a Star, taken from the bot's own Star balance. Telegram notes that bots with increased limits are charged only for messages that are broadcast successfully.

Read the entry requirement before you get excited. To enable the feature a bot must hold at least one hundred thousand Stars on its balance and have at least one hundred thousand monthly active users. That is not a tier for a growing business, it is a tier for an application with a national footprint. Know the number precisely so that nobody sells you a paid broadcast plan you cannot qualify for.

What the bot road cannot do

It cannot start conversations. It cannot rescue a list of strangers. It cannot pretend to be a person. If your plan depends on reaching people who have never interacted with you, this section is background and the next one is your actual situation.

Route two: senders that run on user accounts

This is the category most people mean when they type the search phrase. A tool logs into one or more real Telegram accounts and sends from them, one message at a time, to whoever is on the list.

What you are actually buying

You are buying scheduling and bookkeeping. The sending itself is the same call the app makes. What a good tool adds is the discipline a person cannot keep for eight hours: an even gap between messages, a queue that survives a restart, a different account picking up when one is resting, and a written record of which recipient got which variant.

What you are not buying is permission. The account is subject to exactly the rules a human account is subject to, which is the section further down that decides whether any of this works.

Hosted or installed

Hosted means the sending happens on the seller's servers and you manage it from a browser. Installed means the sending happens on your machine, from your address, and stops when your machine sleeps. Neither is automatically better, but they fail differently. A hosted tool keeps running while you close the laptop, and it also means someone else holds your logins. An installed tool keeps the logins local, and ties every campaign to one computer being awake, on one connection, which is a pattern in itself.

Where multiple accounts help, and where they do not

The pitch for many accounts is capacity. The real benefit is isolation. Ten accounts do not multiply how much unwanted messaging Telegram tolerates. They mean one restriction does not stop everything, and they let you find out which message and which list segment cause trouble, because the accounts differ.

Here is the trap, from our own campaigns: the same message from ten accounts produces ten restrictions, just later. Accounts spread the consequence, they do not remove the cause. If your reply rate is near zero, adding accounts buys you a longer, more expensive version of the same result.

Route three: browser extensions and web client scripts

These run inside Telegram Web. You install an extension, open the web client, and it clicks through your list on your behalf.

Why they are popular

They are cheap or free, they need no phone number handling that you can see, and they feel safe because the sending happens in a window you are looking at. There is no server, no session file to hand over, and the setup takes two minutes.

What actually happens

The extension has whatever access your logged in browser session has, which is the same as full access to your account. It reads your chats because it has to. Whether it does anything else with them is a question the code answers, and you are usually not reading the code.

Practically, these tools are also the slowest and the most fragile. They send at the speed the web interface will accept, they break when the interface changes, and they cannot pace anything while the tab is closed. As a way to send fifty messages to a saved list they are fine. As the engine for a real campaign they are a browser tab pretending to be infrastructure.

The one thing worth checking

Look at what permissions the extension requests. An extension that asks to read and change data on all sites is asking for more than a Telegram sender needs. This is a two minute check that most people skip, and it is the difference between a narrow tool and a wide open door.

Route four: desktop senders you download

The classic shape: an installer, a licence key, a window with a list box and a send button. This category is where the free searches land, and it deserves a careful, unexcited look.

What is good about them

The logins stay on your machine. There is no monthly bill in some cases. For a one time job with a modest list they can be the right amount of tool, and some of them are honest software written by one person who understands the protocol.

What to check before you run one

You are giving an unknown executable your phone number and the login code that comes with it. That is complete account access, not partial. Once a program has completed that login, it holds a session that keeps working even if you change your password later, because a password governs new logins and does nothing to sessions that already exist.

Two habits make this survivable. First, log the tool in on an account you can afford to lose, never the account that owns your channels and groups. Second, open the active sessions list in your Telegram settings after the first run, look at what appeared, and end anything you do not recognise. If the tool's session shows up from a country you have never visited, that is your answer about where the software actually runs.

The category's honest limitation

A desktop sender sends from one address, on one machine, in whatever pattern the software was written to use. It cannot rest accounts across a day, it cannot resume a queue after a crash it did not survive, and it usually cannot tell you why a message failed, only that it did. That is a real ceiling, and it is the reason serious operations move off this category rather than upgrading inside it.

Route five: paying somebody else to send

A managed service takes your message and your target definition and does the sending on its own accounts. You get a report at the end.

When it is the right call

When you want the result and not the machine. There is nothing to install, no accounts to buy, no restriction to appeal, and the operational risk sits with the provider. For a first test of whether a market responds at all, this is often the cheapest way to find out, because a failed experiment costs you the fee and nothing else.

What to insist on

Ask what the report contains. A number of sends is not a result. Delivered, failed, blocked and replied are a result, and a provider running the campaign properly has all four. Ask who owns the accounts and what happens to your message text afterwards. Ask what happens if delivery stalls halfway, and get the answer before you pay rather than after.

The trade you are making

You are trading control and learning for convenience. You will not learn which message variant worked on which segment unless the provider tells you, and many do not, because the campaign is their product and the learning is a by-product they never packaged. If you plan to run outreach continuously, own the machine. If you are testing whether a market exists, rent it.

The ceiling no sender can raise

This is the section that decides whether any of the five routes works for you, and it has nothing to do with software. Telegram publishes a short page for people whose accounts have been limited, and it is the most useful document in this entire subject.

Limits come from people, then from moderators

Telegram's own explanation is that when users press the report button in a chat, those messages go to a team of moderators for review, and if the moderators agree, the account becomes limited temporarily. The trigger is a human objection and the decision is a human decision. There is no throttle a tool can tune, because the input is not traffic volume, it is annoyance.

That reframes the whole category. A sender does not control the ceiling. The message controls the ceiling, and the list controls the message's chance of being welcome.

What being limited actually means

Telegram is specific about this, and it is less total than people assume. A limited account can still message people who have its number saved as a contact, and it can always reply to anyone who wrote first. What it loses is the ability to open conversations with strangers, which is precisely the ability an outreach campaign runs on.

The duration scales with history. Telegram says a first offence, for someone who is not an industrial scale spammer, usually means a few days. Repeated offences mean longer periods. Keep going and the ability can be lost permanently.

What people report you for

Telegram states that for private messages it does not really matter what you send, as long as the receiver finds it unwelcome, and lists what people do mind: unsolicited advertisements, links, invite links to groups or channels, random photos, and above all anything related to commerce or online popularity. It also says plainly that username search is not a tool for making new friends.

Read that list again next to your draft. If your first message is an advertisement with a link, you are not near the line, you are on the list.

The new number trap

Telegram acknowledges that some numbers trigger a harsh response before the account has sent anything, either because of what a previous owner did with that number or because it is a virtual or voice over internet number. If you buy accounts and half of them are limited on day one, this is usually why, and it is not something your sender did.

The spam restriction guide covers the appeal path through Telegram's own bot, which is the only route that exists.

And the rule everybody skips

Telegram's terms of service open with a short list of things you agree not to do, and the first item is using the service to send spam or scam users. Any tool marketed as a way around that is marketed against the platform's own terms, and the account paying the price is yours, not the vendor's.

What free actually costs

The autocomplete for this topic is full of the word free, so it deserves a direct answer rather than a lecture.

Free is real, and it is narrow

A free tool that sends messages from your account genuinely exists and genuinely works. The code is not hard. If you have one account, a list of two hundred people who know you, and an afternoon, a free script is the correct choice and paying for a platform would be a waste.

Where the cost shows up instead

The price of free is paid in four places, and none of them appear on the download page.

The account. A free tool almost never paces properly, because pacing is the boring part nobody writes for fun. The bill arrives as a restriction on the account you used.

The record. Most free tools tell you they finished. They do not tell you that ninety of five hundred failed, that forty of those were privacy refusals and fifty were dead usernames, which is the only information that improves the next run.

Your time. Free tools break when Telegram changes something, and the person fixing it is you, on the day you planned to run a campaign.

The unknown. A downloaded executable that takes a login code is a complete account handover to a stranger. Sometimes the stranger is a decent developer. Sometimes the account is resold. You have no way to tell in advance, which is the actual cost.

The honest rule

Free is right when the job is small, the account is disposable and the list is warm. Paid earns its price when the job repeats, because everything you are paying for is repetition: pacing that holds, records that accumulate, and an account pool that survives a bad week.

Reading a feature list honestly

Every sender's page lists the same five or six capabilities. Here is what each one is actually worth, in the order most people rank them wrong.

Account rotation

Sold as capacity, worth buying as isolation. Rotation matters because it contains damage and because it lets you compare, not because it multiplies your allowance. A tool that rotates but sends the same text from every account has kept the mechanism and lost the point.

Message variation

Also called spinning: the tool rewrites parts of each message so no two are identical. This is worth more than rotation and costs less. Identical text arriving from ten accounts is the single most recognisable pattern in this entire field, to recipients as much as to anyone else. Variation is also the cheapest thing to fix, because it changes nothing about your infrastructure.

Delays and scheduling

Look for a range rather than a number. A fixed twenty second gap is a metronome, and a metronome is a signature. Look also for whether the tool respects hours of day, because a first message at three in the morning in the recipient's time zone is answered by nobody and reported by some.

Warmup

Some tools can have accounts talk to each other or read channels for days before a campaign, so a new account has some history. This helps, and it is slow by nature. Treat any tool that offers instant warmup with suspicion, because the thing being built is age and age cannot be hurried.

Connection handling

If a tool supports one address for all accounts, that is one signal shared by everything you own. Separate connections per account is the feature to look for, and it is worth more than most of the items above it on a marketing page. Our own connection guide goes through the types and what each is worth.

Reporting

The least exciting feature and the one that decides whether your second campaign is better than your first. Sent, failed, blocked, replied, per recipient, exportable. If a demo cannot show you that screen, ask why.

The list decides more than the sender

Two campaigns with the same tool, the same accounts and the same message can end a week apart, one with replies and one with restrictions. The variable is the list.

What a member list really contains

Group member counts are not reachable people. Measured on our own runs: a group displaying 10,339 members returned three reachable accounts, and all three were bots. Another group of 372 members yielded 369 usernames; one of 677 yielded 474. Across nine groups, 2,180 raw members reduced to 2,145 unique people, which means the groups barely overlapped and nothing was saved by merging them.

The lesson is not that collecting is pointless. It is that the visible number and the usable number are different numbers, and any plan built on the visible one is wrong before it starts. The group collection guide covers the mechanics.

Relevance beats size, measurably

One of our own customers had a main search term that returned nothing usable, while fifteen neighbouring terms returned thirty five groups. The instinct is to broaden until the list is big. The correct move is to narrow until the list is right, because every irrelevant recipient is a person with a report button and no reason to be polite.

Duplicates and dead usernames

A person in three of your groups is one person and three messages unless the tool removes duplicates. A username that no longer exists is a failure your report should label honestly rather than counting as a send. If a sender cannot deduplicate across sources, you will annoy your best prospects most, because the most engaged people are in the most groups.

Logins, sessions and what you hand over

Every tool on the account side needs a login, and this is where most of the real risk in the category lives.

A session is not a password

When a tool logs in, it holds a session. Changing your password later governs new logins and does not end sessions that already exist. Ending a session is a separate action, in the devices list inside Telegram's own settings, and it is the only thing that actually revokes access.

Session files are portable

A session file is a small database, and it is not encrypted by default. We opened an empty one to check: five tables, no password. Anyone holding that file holds the account, on any machine. A used file can also carry the previous owner's cached contact entries, which is worth knowing if you ever buy accounts with sessions attached. The session file comparison covers the formats.

Three habits that cost nothing

Use a separate account for sending, never the one that owns your channels. Review the active sessions list after any new tool logs in, and end anything unfamiliar. If a tool wants a two step verification password as well as a code, understand that you have handed over the ability to keep you out, not just the ability to get in.

How to judge a sender before you pay

Twelve questions. Every one of them has a correct answer that a good vendor gives immediately and a vague one that tells you as much.

The four that eliminate

Does it need a bot token or a phone number? Can it message people who have never contacted me, and if the answer is yes, on which side of the API? Where does the sending run, my machine or yours? Who else can see my logins?

The four that separate good from adequate

Can each account use a separate connection? Is the delay a range or a fixed number? Does the message vary per recipient, and can I see the variants? Does the queue survive a restart, and what happens to the messages that were in flight?

The four about what happens afterwards

What does the report show per recipient? Does it distinguish a privacy refusal from a dead username from a block? Can I export it? And when an account gets restricted, does the tool stop that account, keep the rest running and tell me which message was in flight when it happened?

The question that is not on the list

Nobody should ask a vendor how many messages a day is safe, because there is no honest answer. Telegram publishes no such number, the ceiling moves with recipient reaction, and any vendor who names a confident figure has just told you they will guess about other things too. From our own sending, a fresh account handles roughly forty to fifty new conversations a day before trouble starts, and an account with two or three months of ordinary history handles roughly eighty to a hundred. Those are our numbers, on our accounts, with our messages. Treat them as a starting point to measure from, not a promise.

What good numbers look like

Once a sender is chosen, four numbers tell you whether the campaign is healthy, and none of them is volume.

Delivery rate

The share of attempts that arrived. A low rate this early usually means the list is stale rather than the tool is bad. Dead usernames and privacy refusals both show here, which is why they need separate labels.

Reply rate

The only number that predicts revenue. It is also the number that predicts safety, because a message people answer is a message people do not report. If replies are near zero, adding accounts and raising volume makes the outcome worse, not slower.

Block rate

Your early warning. Blocks rise before restrictions arrive, and they rise on a specific message or a specific segment. A report that cannot show blocks by message variant is hiding the one signal that would have saved the account.

Restriction rate per account

How many of your accounts are limited at any time, and how quickly they come back. This is the number that tells you whether your pacing is real or decorative. Our own guide to avoiding restrictions in cold outreach goes deeper into the pattern.

One measurement warning

If your message carries three links and you only measure clicks on one, your click rate is a third of the truth. We got this wrong on our own campaign before we caught it, and it is the easiest way to conclude a good message failed.

Where Floqal Mass DMs sits in this list

We build one of these tools, so here is the honest placement rather than a pitch.

Mass DMs is on the account side, hosted. It runs on our servers, so campaigns continue when you close the tab, and you manage them from a browser with nothing installed. Accounts rotate, each one can use its own connection, delays are ranges rather than fixed gaps, messages vary per recipient, and the report is per recipient with sent, failed, blocked and replied separated rather than merged into a single total.

What it does not do: it does not raise Telegram's ceiling, because nothing does. It will not make an unwelcome message welcome, and it will not keep an account safe if the list is wrong. If your list is your own subscribers, we would point you at the bot road above instead, because it is cheaper and calmer and you do not need us for it.

Tiers and what each includes are on the pricing page, and a demo opens from inside the dashboard after a short qualification step, which exists so that the people who get one are the people who will actually use it.

Choosing, in one page

If your list is people who already contacted you, use the Bot API and stop reading. It is free up to thirty messages a second and nothing can be restricted.

If your list is strangers and the job is one time and small, a free script on a disposable account is proportionate. Accept that you will not learn much from it.

If your list is strangers and the job repeats, buy the account side, hosted, with per account connections, ranged delays, per recipient reporting and message variation. Those five features are the difference between a campaign you can improve and a campaign you can only repeat.

If you want the outcome without the machine, pay someone to run it, and insist the report separates delivered, failed, blocked and replied.

And whichever you choose, remember what the platform's own page says: the limit arrives because a person objected, not because a counter filled up. The tool is the smaller half of this decision. The message and the list are the larger half, and they are the half you control completely.

The failures you will see, and what each one means

A good report separates these. A weak one calls all of them failed, which is why two campaigns with identical failure counts can need opposite fixes. Here is how to read them.

The recipient does not allow messages from strangers

A privacy setting, chosen by the person, and nothing to do with your account's health. There is no workaround and no reason to want one. A list with many of these is not a broken list, it is a list of people who have decided in advance, and the correct response is to accept the number and stop paying for those rows.

The username no longer exists

Either the person changed it, deleted the account, or your list is old. Treat a rising count here as a date stamp on your data rather than a fault in the sender. One caution from our own logs: a sending account that has been restricted can also fail to resolve usernames that are perfectly valid, and a careless tool labels that as the recipient being gone. If a name you can open by hand comes back as missing, suspect the sender's account, not the recipient.

The recipient blocked you

The clearest signal in the whole set, and the one to watch by message variant rather than in total. Blocks precede restrictions. If one variant collects them and another does not, you have just been handed the answer for free.

Too many messages, wait

A pacing response, and the only failure on this list that is genuinely about speed. A tool that treats it as a hard error and retries immediately turns a pause into a pattern. The correct behaviour is to wait out the interval on that account and let the others carry on.

The account cannot start new conversations

This is the restriction described earlier, and it is the one that ends campaigns. Note that it does not stop replies, so a restricted account can still finish conversations already in progress. Move the queue to another account, appeal through Telegram's own bot, and treat the message that was in flight as the first suspect.

Before you buy: the smallest sensible test

Whatever you pick, do not start with the full list. Run one account, one hundred recipients, two message variants, and normal pacing. It costs a day and it answers questions that no vendor page can.

You will learn the delivery rate on your real data, the split between privacy refusals and dead entries, whether either variant draws blocks, and whether the account is still healthy at the end. If replies land, scale the message that produced them. If none do, the message is wrong, and running the same words at ten times the volume converts a quiet failure into an expensive one.

A demo is the cheapest version of this test, which is why our own runs on a real account with the full feature set for a day rather than a stripped down preview. What matters is not what a tool looks like, it is what your list does when it meets it. Support scope and what we cover if something goes wrong are on the warranty page.

Frequently Asked Questions

Can a Telegram bot send bulk messages to people who never started it?

No. A bot can message people who already have a chat with it, which means they tapped start or joined a group where the bot is a member. This is the design of the Bot API rather than a limit you can lift, so any tool promising bot based cold outreach is describing something the platform does not do.

How many messages per second can a bot send?

Telegram's own bot documentation gives three numbers: no more than one message per second in a single chat, no more than twenty messages per minute in a group, and no more than about thirty messages per second for bulk notifications. Paid broadcasts raise the bulk figure to a thousand per second at a tenth of a Star per message above the free thirty.

What does a paid broadcast actually require?

To enable it, Telegram states that a bot must hold at least one hundred thousand Stars on its balance and have at least one hundred thousand monthly active users. Charges apply only to messages that broadcast successfully. For most businesses this tier is out of reach, which is worth knowing before a vendor quotes it as an option.

Is a free Telegram bulk message sender safe to use?

The sending itself is the same call any tool makes, so the risk is not in the sending. It is in what you hand over and what the tool leaves out. A downloaded program that takes your login code holds complete account access, and free tools rarely pace properly or report failures in a way you can learn from. For a small job on a disposable account, free is proportionate. For anything repeating, the cost reappears as restrictions and lost information.

How many messages a day can one account send safely?

Telegram publishes no such number, and any vendor who names one confidently is guessing. The limit follows recipient reaction, not a counter. From our own sending, a fresh account handles roughly forty to fifty new conversations a day and an account with a few months of ordinary history handles roughly eighty to a hundred, but those are our figures with our messages and should be treated as a place to start measuring.

Will more accounts let me send more?

More accounts spread the consequence, they do not raise the allowance. The same message from ten accounts produces ten restrictions, just later. Accounts are worth buying for isolation and for comparison between segments, not as a multiplier on volume.

What happens when an account gets limited?

Telegram explains that a limited account can still message people who have its number saved as a contact and can always reply to anyone who wrote first. What it loses is the ability to start conversations with strangers. A first restriction usually lasts a few days, repeated ones last longer, and persistent behaviour can remove the ability permanently.

Why was a brand new account limited before I sent anything?

Telegram acknowledges that some numbers trigger a harsh response on their own, either because of what a previous owner did with that number or because it is a virtual or voice over internet number. This is a common experience with bought accounts, and the appeal route is Telegram's own spam bot rather than the seller.

Do browser extensions get accounts restricted more often?

Not because they are extensions. They get accounts restricted because they usually send with no pacing control and no failure handling, and because they stop the moment the tab closes, which pushes people to run everything in one burst. The extension is also holding full access to your logged in session, which is a separate question worth answering before installing one.

Is bulk messaging against Telegram's rules?

Telegram's terms of service list what you agree not to do, and the first item is using the service to send spam or scam users. Messaging people who expect to hear from you is ordinary use. Messaging strangers with advertisements is the exact behaviour the reporting system exists to catch, and Telegram names unsolicited advertisements and links among the things people mind most.

What is the difference between a bulk sender and a mass DM tool?

In practice, nothing: they are two names for the same category, and vendors pick whichever phrase their market searches for. What actually differs between products is the side of the API they use and the quality of their pacing and reporting, which is why comparing them by name gets you nowhere.

Which numbers tell me a campaign is working?

Reply rate first, because it predicts both revenue and safety. Then delivery rate to judge the list, block rate as an early warning before restrictions arrive, and restriction rate per account to judge your pacing. Volume is not on the list, and treating it as the headline is how campaigns end early.