Comparison — AI SDR
An AI SDR wins on volume. European outbound is not a volume problem.
An AI SDR does not hold a presumed-consent file, sit a mock call in Finnish, or scrub the Italian opt-out register every 15 to 30 days. In Europe, those three things are the job.
We automate parts of our own pipeline. This page says which parts.
Where automation wins. Including where it beats us.
11 jobs make up an outbound programme in Europe. Automation wins six of them outright, and we run those six on automated tooling ourselves. One is split. The remaining four are the reason this page exists.
| Job | Automated tooling | A person in market | What we run |
|---|---|---|---|
| Building the target list | Assembles, dedupes and de-conflicts thousands of accounts in an afternoon. | The same list, several days later. | Automated |
| Enriching and verifying a record | Checks a role, a domain and a work address against several sources, on every record, without getting bored at record 900. | Slower, and no more accurate. | Automated |
| Sequencing mechanics | Sends at the right local hour, stops on reply, never skips a step and never sends twice. | This is the part people are worst at. | Automated |
| First-touch email at scale in English-speaking markets | Where the lawful basis is settled and the market works in English, the constraint is how many accounts you reach. That is what this is for. | A person writing the same email is the same email, later. | Automated |
| Cost per touch | Cents. | Not cents. Nowhere near cents. | Automated |
| Logging, recording capture and reporting | Every send, every reply and every recording, timestamped, without anyone remembering to write it down. | Worse. Every time, in every company either of us has worked in. | Automated |
| Deciding a German business may lawfully be called | Stores the evidence and the date it was gathered. | Decides whether the evidence is concrete enough, writes down why, and is willing to explain that to a regulator. | A person |
| The first call, in the prospect's first language | Dials, speaks, and gets the grammar right. | Sells in the register that market actually uses, and hears the objection underneath the objection. | A person |
| The objection that comes back mid-call | Answers what was said. | Answers what was meant, and knows which of the two the prospect wanted. | A person |
| Reading whether the market answers at all | Produces a reply rate. | Names the three reasons the market said no, and which of them are fixable. | A person |
| Evidencing a meeting you dispute | Attaches the recording and the timestamps. | Reads the recording against the rubric and accepts it when the meeting fails. | Both |
Those six are not concessions we are making to sound even-handed. They are the six jobs we automate, and if the only thing you need is those six, buy the automation. You would be buying the right thing.
Automate the work that repeats. Put a person on the work that has to be defended.
Every job automation won in that table has one thing in common. Being wrong costs you a deleted email.
Every job it lost has the other thing in common. Being wrong costs you a letter from a regulator, or the one meeting in that market that was worth having.
That is the whole argument. We are not going to tell you the tools do not work — we would be arguing against our own pipeline, and we would be wrong within a year.
When to buy the automation, and when to put a person on it.
This is the test we run before we quote a market. Two of the conditions on the left describe markets we would tell you not to hire us for.
Buy the automation when
- A1Your first touch is email, and the lawful basis for that email in that market is settled.
- A2The market works in English, or in a language somebody on your team already answers in.
- A3The constraint is how many accounts you can reach, not which ones will take the call.
- A4Being wrong costs you a deletion, not a filing.
- A5Someone on your side can answer the reply within the working day.
Put a person in the market on it when
- B1The lawful basis is a judgement about one specific company, and somebody has to write it down before the first contact.
- B2The first touch has to be a call, because email is not available to you in that market.
- B3The reply comes back in a language nobody on your team reads, and it needs answering the same morning.
- B4The market's norm is relationship-first, and a US-direct opening ends the conversation inside 20 seconds.
- B5Being wrong is expensive — a suppression-list failure, a call to a registered number, or a meeting you cannot evidence.
Most European programmes need both, in that order: automation builds and verifies the list, a person makes the contact that carries the risk. The expensive mistake is not choosing automation. It is choosing it for line B1.
Some obligations are judgements. A judgement has an author.
Some of what European outbound asks for is mechanical: screen this list, honour that register, re-run it on Tuesday. A system does all of that better than a person.
Some of it is a judgement about one company, written down before the first contact, by somebody who would have to explain it. That is not a data field. It is a person.
34 markets. The obligations are not the same job.
Until counsel has signed off a market, we publish the kinds of obligation, not a legal position for that market. The legal position for all 34 markets · How a meeting is evidenced.
| Obligation | What automation does | What it cannot |
|---|---|---|
| A judgement about one company, written down before contact | Stores the evidence and the date it was gathered. | Decide whether the evidence is concrete enough, and be prepared to say why. |
| Screening against a live opt-out register | Holds a snapshot of the list at the moment it was loaded. | Re-scrub at short intervals and refuse the dial when the register has moved. |
| Self-identification and an objection route, on the call | Can play a recorded opener. | Carry both, in the market language, and change course when the prospect objects. |
| A defensible file for the contact that was made | Logs the send. | Author the file a regulator would accept. |
Translation clears grammar. It does not clear register.
A tool can produce a grammatically correct Finnish sentence. It cannot sit a mock call in Finnish, hear that the objection is about implementation rather than price, and answer it the same morning.
What a person clears before they touch your market
Certification is sequential and per market. Passing for one market certifies nobody for another. The certification standard in full.
| # | Gate | What it is | Pass condition | If they fail |
|---|---|---|---|---|
| 1 | Mock call | A live call in the market's language, run by someone who has sold into that market. Scored against the same six criteria we score real meetings against. | They book a meeting that would pass the rubric, and they can say which line of the call met each criterion. | Re-sit. No campaign in the meantime. |
| 2 | Native-language screen | Held entirely in the market language, by a native speaker. Not a language certificate — business register, sector vocabulary, and holding an objection without switching to English. | The screener would take the call from them as a buyer in that market. | No pass, no campaign in that market. There is no English fallback in a market we sold as native-language. |
| 3 | Compliance test | A written test on the rules for the markets they'll be calling into: what's permitted, what needs consent, what's prohibited, what must be logged, and what to do the moment someone objects. | Every question on prohibitions and on objection handling answered correctly. Those are the ones that create liability. | Re-sit after retraining. No campaign in that market until they pass. |
| 4 | Product comprehension | They explain your product back — what it does, who it's for, and what it doesn't do — and we check that against the material you gave us at kickoff. | They can answer the three objections your own team hears most, and they can say plainly where the product isn't a fit. | Back to the material. Run once per campaign, not once per career. |
Certification is per market and per campaign. Passing for Germany certifies nobody for France.
This is a hiring standard, not a forecast. It tells you what a rep passed before they started. It doesn't tell you what they'll book in month one, and we're not going to pretend it does.
If your market works in English and your constraint is volume, buy the automation. Not us.
We are not the right supplier for a campaign where the lawful basis is settled, the language is one your team already answers in, and the only real question is how many accounts you can reach this quarter. Automation is faster and cheaper at exactly that, and we would be selling you a person to do a machine's job.
We are the right supplier when the first contact is the part that carries the risk.