How to get past gatekeepers in B2B sales calls

A gatekeeper is not a wall. They are a routing function with a very fast classifier, and the question they are answering is not "should this person be allowed through" but "does this call belong to the person they asked for". Everything that works follows from that. You give them what a router needs — a name, a person, a one-line reason in business language — and you ask them to route. Everything that fails is some form of withholding that information, which forces them to classify you as a pitch, because a pitch is the only thing that behaves that way.

What a gatekeeper is actually deciding in the first eight seconds

Put yourself on their side of the desk. They handle a large volume of calls, most of which are either routine business or attempts to reach someone who does not want to be reached. They have no way to evaluate your product and no interest in trying. They are running a much simpler test with three inputs.

Do I know who this is? A name and a company, stated clearly, without hurry. A mumbled name is not humility, it is the single most reliable marker of a call the caller does not expect to be welcome.

Do they know who they want? Asking for a named person is the strongest single signal available. Asking for "whoever handles marketing" tells them you have no relationship and no research, which is exactly the profile they are there to filter.

Can I repeat the reason to that person accurately? This is the one most callers miss. The gatekeeper is not evaluating your reason on its merits. They are checking whether they can say it out loud to their colleague without looking foolish. A reason that is specific, short, and framed in the language of the business passes. A reason that is a value proposition does not, because they cannot repeat it without sounding like they have been sold to.

Eight seconds, three inputs. Nothing about persistence, charm or technique changes the test.

Why "getting past" is the wrong objective

The metaphor causes bad behaviour. If the objective is to get past a person, then tricks are on the table, and tricks are where calling programmes damage themselves.

Reframe the objective as: get the call routed correctly, or get the routing corrected. Under that framing, a call that ends with "she does not handle that, you want Anand in operations" is a success rather than a failure, and it is a success you can only get by being straightforward, because a gatekeeper who has classified you as a pitch will not volunteer a colleague's name.

The reframing also fixes your metrics. If success is "reached the target", most of your switchboard calls fail, and the programme looks pointless. If success is "the record is now correct or the call is routed", a large share succeed, and you start to see the actual value of the channel — which for most B2B teams is data correction as much as it is access.

The switchboard has shrunk, and most gatekeeper advice is about a smaller channel than it was

Two structural changes have narrowed the classic gatekeeper scenario.

Direct dials and mobile numbers have absorbed a large share of business calling, so a good data source often puts you straight through to a person with no intermediary at all. The gatekeeper conversation is now a fallback rather than the default path.

And buyer behaviour has moved away from seller contact generally. Gartner reported on 9 March 2026 that 67 percent of B2B buyers prefer a rep-free experience. That figure is about the buying journey rather than about receptionists specifically, but it points at the same thing: the number of unsolicited conversations any given buyer is willing to have is falling, and the screening function is now distributed across a receptionist, an executive assistant, a spam label on the caller ID, a filter on the mailbox and the buyer's own habits.

The practical consequence is not "stop calling switchboards". It is: do not build a calling programme whose primary access strategy is a switchboard, and do use the switchboard for the thing it is uniquely good at, which is telling you who actually owns the problem inside that building. No data vendor knows that reliably, and the person on the front desk does.

The three things you have to supply, in order

The script is short because everything in it is load-bearing.

One: your name and your company, at normal speed. "Hi, this is Sam Rao from Autocloz." Not fast, not apologetic. A gatekeeper's ear is calibrated for the hurried version.

Two: the person, by name, as a request rather than a demand. "I am hoping to reach Priya Nair." Using the full name is a small signal that you have done work. Using only a first name reads as feigned familiarity, which is one of the patterns they screen for.

Three: the reason, unprompted, in one sentence of business language. Do not wait to be asked. Volunteering the reason removes the moment where they have to extract it from you, and the extraction is where calls die. "I am calling about the two SDR roles they posted last week — I have a question about how the team is covering ramp."

Then ask for help, explicitly. "Is she the right person for that, or should I be speaking to somebody else?" That question does two things at once. It hands them the routing decision, which is theirs anyway, and it opens the door to the most valuable answer in the whole call, which is a different name.

Everything else — tone, pace, politeness — matters, and none of it substitutes for those three inputs. A charming caller who withholds the reason gets screened. A flat caller who supplies all three gets routed.

Scripts for the four routes a gatekeeper can send you down

There are only four outcomes, and each has a correct next move.

Route one: transfer. They put you through. Your cold-call script starts now, and it starts with the same reason you gave the gatekeeper, because the gatekeeper has probably already said it and contradicting yourself is fatal. The structure of what comes next is in cold calling scripts that book meetings.

Route two: message. "She's in a meeting, can I take a message?" Take it seriously. A message that gets written down accurately is a warm introduction from a colleague. Give them your name, your company, one sentence of reason, and a number, spoken slowly and repeated once. Then ask the useful question: "When is a good time to try her directly?" That answer is worth more than the message, because it converts a random dial into a scheduled one.

Route three: redirect. "That's not her, that's Anand." This is the best outcome after a transfer, and frequently better than a transfer to the wrong person. Get the full name and, if offered, the extension or the email. Then update the record immediately — a redirect you do not write down is a redirect you will need again in six weeks.

Route four: refusal. "We don't take sales calls." Accept it in the first sentence, without negotiation. Then one question, once: "Understood. If I wanted to reach the person who owns this, what is the right way — email, the website form?" If that gets a refusal too, thank them and end the call. A second attempt to negotiate the refusal is the point at which a routine screening turns into a complaint.

The highest-value ask is not the transfer

Reps optimise for the transfer because it feels like winning. The redirect is usually worth more.

A transfer to a person who turns out not to own the problem costs you a conversation and teaches you nothing. A redirect gives you a corrected record that improves every future touch on that account across every channel — the email you send next week, the LinkedIn request, the second dial. And the redirect is only available to callers who asked for help rather than demanded access, which is the practical reason honesty outperforms technique here rather than merely being preferable.

So build the ask in explicitly. After you have supplied the three inputs, ask "is she the right person for this?" before you ask to be put through. It reorders the call so that the routing question comes first, which is the question the gatekeeper is best placed to answer and most willing to answer.

Autocloz's free plan covers 5 users and 10 mailboxes, with the dialler, the sequence builder and the lead record in one login rather than three tools — start free if you want the redirect you just collected to update the same record your email sequence reads.

"Do not call here again" is a legal event, not a bad call

This is the part that turns a soft-skills topic into a compliance one, and it is routinely mishandled.

In the United States, a company-specific do-not-call request is not a preference. Under 47 CFR 64.1200(d)(3), persons or entities making such calls "must honor a residential subscriber's do-not-call request within a reasonable time from the date such request is made", and that period "may not exceed ten (10) business days from the receipt of such request". Under 64.1200(d)(6), "a do-not-call request must be honored for 5 years from the time the request is made". Paragraph (d)(1) requires that anyone making calls for telemarketing purposes has "a written policy, available upon demand, for maintaining a do-not-call list".

Ten business days, five years, and a written policy. Those three numbers are the whole operational requirement, and none of them is satisfied by a rep making a mental note.

Two things follow for how you handle a refusal. First, record it the moment the call ends, against the number, with the date. Second, understand what your suppression actually covers. In Autocloz a do-not-contact entry is keyed to the identifier rather than to the person, and the wildcard channel covers every channel for that identifier. So suppressing the switchboard number stops calls, SMS and WhatsApp to that number and does nothing at all to the email addresses at that company. If the refusal was on behalf of the organisation rather than that one line, you have to suppress the other identifiers deliberately.

The two federal statutes this sits inside, the three separate do-not-call lists, and what actually changed in 2025 are covered properly in TCPA compliance for cold calling. Treat all of the above as a description of published rules rather than as advice on your programme, and take advice on your own facts.

Recording the attempt so the second call is not the first call again

Most of the value of a gatekeeper conversation is destroyed in the ten seconds after it ends, when nothing is written down.

Four things are worth capturing every time, and all four fit in a lead's custom fields:

• The routing answer. Who owns this, according to the person who routes calls for a living.
• The best time. If they told you when to try, that is a scheduling instruction, not trivia.
• The screening posture. Whether this front desk transfers, takes messages, or refuses. It determines whether calling this account again is worth anyone's time.
• The exact refusal, if there was one. Wording and date, because it is the record you would produce if asked.

In Autocloz a PATCH to a lead's custom fields shallow-merges — writing one key updates that key and leaves the rest of the map intact, and sending a key as null removes it — so the rep who adds a routing answer today does not wipe the screening note from last month. That merge behaviour is what makes incremental account notes safe.

One convention trap to design around before you build a tagging scheme on top: a tag is stored exactly as typed, but the ?tag= filter lower-cases the value before it matches, and the underlying match is case-sensitive. A tag saved as "Gatekeeper-Refused" cannot be found through that filter. Use lower-case tags and enforce it in your process.

And warm the account before the next dial rather than repeating the same cold attempt. An email and a LinkedIn touch ahead of a call mean the name is recognised when the gatekeeper repeats it, which measurably changes how the reason is received — the sequencing rules and the per-channel caps that govern that are in sales cadence best practices, and the mechanics of getting the resulting conversation into a calendar slot are in the B2B appointment setting guide. The dialler surface is where the disposition and the note get written without leaving the call, and the multichannel sequence builder is where the warming touches live.

The tactics that fail, and why each fails structurally

These are not fails-sometimes tactics. They fail for reasons built into the situation.

Refusing to say what the call is about. It forces the gatekeeper to classify with one input missing, and the default classification for a missing reason is "sales call". You have chosen the worst branch on purpose.

Implying an existing relationship. "She's expecting my call" is checkable in about four seconds, and being caught converts a neutral screening into an account-level block.

Manufactured urgency. Someone who screens calls all day has heard every urgent variant. Urgency that is not backed by a fact reads as pressure, and pressure is the clearest pitch marker there is.

Calling repeatedly at different times to catch a different person. It works occasionally and it produces a front desk that recognises your number, which is worse than the one call you avoided.

Going around them to a direct dial after a refusal. If the refusal was personal preference, this is rude. If it was a company-specific do-not-call request, it is a compliance failure with a five-year memory.

Treating the gatekeeper as an obstacle in your tone. They can hear it. Everyone can hear it. It is the fastest route to route four.

What this playbook cannot do, and what Autocloz does not do

No script gets you to someone who does not want to be reached. A gatekeeper who is following an explicit instruction from their executive is not a puzzle to solve. The correct response is to change channel, change the person, or accept the answer.

None of this fixes a bad reason. If the reason you are calling is not specific to this company, the gatekeeper is right to screen you and the transfer would have wasted the executive's time as well. Relevance is upstream of access.

Autocloz does not check consent before a call. The call dispatcher checks the do-not-contact list for that number and the quiet-hours window on the sending account, and stops or defers on either. It does not consult the per-contact consent ledger — opt-in enforcement is implemented for WhatsApp only and is off by default there. Consent for a calling programme is yours to establish and record.

Quiet hours use the sending account's clock, not the recipient's. Autocloz resolves the window as a campaign override, then the workspace default for the channel, then the account's own setting, and the time zone comes from the override or the account, defaulting to UTC. The federal rule is written against local time at the called party's location, so a list spanning time zones needs a sender per zone with its own window.

Autocloz does not detect an opt-out spoken on a call. The automatic opt-out detector is deliberately a no-op for calls and LinkedIn, matching carrier keywords on SMS and WhatsApp and a small set of explicit phrases on email. A verbal "do not call here again" is only recorded because a human records it, which is precisely why the ten-business-day clock in the rule above is a process problem rather than a software one.

A do-not-contact entry does not travel across identifiers. Suppressing the phone number does not suppress the email address, and vice versa. If you want a whole account suppressed you have to say so, per identifier.

If you are weighing a CRM with a built-in dialler against a dedicated calling platform bolted onto a separate CRM, the Autocloz and Aircall comparison lays out where each design puts the record.