Cold calling scripts that work: what the data says

On the recorded-call evidence, a cold call script that works does four things: opens with a familiar, human question, states the reason for the call in one line, asks a handful of targeted questions instead of pitching, and slows down when an objection lands. The evidence here is unusually deep. Gong analysed 90,380 cold calls and found the opener "How have you been?" lifted success rates 6.6 times over the 1.5 percent baseline (2018 dataset). Cognism's 2026 report puts the average cold call at 82 seconds, so every line must earn its place. 6 Seven Labs builds AI voice calling agents for Singapore businesses, and this is the script logic we encode into them. Below: what the data supports, a skeleton you can copy, and what changes when software delivers it on every dial.

What do 90,380 recorded cold calls say about openers?

The largest public study of cold call opening lines is Gong's analysis of 90,380 cold calls, built on its 2018 dataset (the page is still maintained in 2026; it remains the largest public opener study we know of). Across all calls, the baseline success rate was 1.5 percent. One opener beat that baseline 6.6 times over: "How have you been?" converted at 10.01 percent.

The lift makes sense once you hear it on a call. The question assumes familiarity and sounds like a person, not a pitch, which buys a few seconds of genuine attention. Even the plain "How are you?" performed well: 5.2 percent, or 3.4 times the baseline. Politeness, it turns out, is measurable.

Stating your reason works too. Openers that included the phrase "The reason for my call is" lifted success 2.1 times over baseline in the same study. People respond to a caller who says plainly why they are calling.

One classic line deserves deletion from every script: "Did I catch you at a bad time?" It converted at 0.9 percent, 40 percent worse than the baseline. It sounds considerate, but it hands the prospect a ready-made exit before you have said anything worth hearing.

The pattern holds in newer data. Cognism's 2026 competitiveness research found top-performing teams open with short, focused intros rather than long warm-ups. Whichever line you choose, the opener has one job: earn the next sentence, not deliver the pitch.

Should you talk more than you listen on a cold call?

In the first minute, yes, and that surprises most people. Gong's cold call statistics (2021 dataset) found successful cold calls run nearly twice as long as unsuccessful ones, that the rep speaks about 55 percent of the time on calls that succeed, and that winning calls carry a longer uninterrupted opening monologue: 53 seconds, against 25 seconds on calls that fail.

The classic listen-more advice is still true, just for a different moment. Gong's separate talk-to-listen research across 326,000 sales calls (updated August 2025) covers sales conversations broadly rather than cold calls specifically, and puts the golden ratio across that wider set at 43:57 talk to listen, while the average rep runs 60:40. On closed-won deals, reps talked 57 percent of the time; on lost deals, 62 percent.

Both findings fit together. A cold call interrupts someone's day, so the caller has to carry the open: who you are, why you are calling, what is in it for them, all inside a minute. Once the prospect engages, the call becomes discovery and the ratio flips towards listening. Cognism's 2026 competitiveness research shows the same shape: top teams avoid long monologues and ask roughly 35 to 45 questions per hour of calling. A 53-second opener is not a monologue problem; ten minutes of pitching is.

Talk ratio is also one of the few numbers worth pulling out of your recordings. We cover which figures deserve a place on your dashboard in outbound calling metrics that matter.

The script skeleton: copy this structure

Here is the structure the data supports, in six blocks. Adapt the words to your business, keep the order. Every technique below traces to a study cited in this post; none of it is folklore.

1. Opener. Greet the prospect by name, introduce yourself, then ask "How have you been?" (10.01 percent success in Gong's 2018 dataset) or the softer "How are you?" (5.2 percent). Never ask whether you caught them at a bad time (0.9 percent).
2. Reason. One sentence beginning "The reason for my call is", the phrasing that lifted success 2.1 times over baseline in the same study. Be specific and honest about why you picked up the phone.
3. Value line. One short sentence on the outcome you deliver for businesses like theirs, stated plainly rather than framed as a request for permission. Keep the whole intro tight; Cognism's 2026 research found top teams open with short, focused intros.
4. Discovery questions. Three or four targeted questions on a first call, drawn from a bank of 11 to 14 built for the full discovery conversation (Gong, 519,000 calls, 2017 dataset). Spread them through the call; the same research found question count stops helping beyond 14.
5. Objection response. Pause first, then ask a clarifying question before you answer, and keep your speaking pace level. The evidence for each of those moves is in the next section.
6. Close. Propose one specific next step with a specific day and time, then confirm it back. If you control your dialling windows, aim at mid-morning: Cognism's 2026 report found 10am to 11am the strongest hour.

One step comes before any of this: the list itself. Before a single number gets dialled in Singapore, run the pre-campaign checks in our guide to PDPA and DNC compliance for outbound calling.

How should your script handle objections?

Slowly, and with a question. Gong studied objection handling across 67,149 sales calls (2019 dataset; it covers sales conversations broadly, not only cold calls) and found top performers differ from average reps in three measurable ways.

• Pause longer. Top performers pause about five times longer after an objection lands; average reps rush to fill the silence.
• Hold your pace. Top performers keep a steady rate of roughly 176 words per minute under pressure; average reps speed up to 188.
• Ask before answering. Top performers respond to objections with clarifying questions 54.3 percent of the time, against 31.0 percent for everyone else.

The script implication is direct: write your objection responses as questions, not comebacks. If a prospect says they already have a provider, the scripted reply is not a rebuttal; it is a question about what is working and what is not. The pause and the pace cannot be written down, though. They are delivery habits, and delivery is exactly where humans wobble.

From script on paper to script on every call

Everything above has a catch: it only works if it happens on every call. In our experience reviewing recorded calls, drift is the default. After a run of rejections the opener gets rushed, the reason mutates, questions get skipped and the pace creeps up, exactly the 188 words per minute the objection research measured. The script on paper and the script on the phone quietly part ways.

That is the case for putting the script in software. An AI voice agent executes the approved structure identically on dial 1 and dial 500: the opener is never skipped, the questions arrive in order, and the pause after an objection is designed in rather than left to nerve. When we build an agent, the script, voice and calling rules are tuned to the business and done for you, and every call is recorded and transcribed, so the script gets reviewed against real conversations instead of memory. Compliance tooling, including a do-not-call list and honoured stop requests, is built in.

Consistency compounds with volume. The average cold call succeeds 2.7 percent of the time on Cognism's 2026 report, so results come from running the same structure across hundreds of dials. List quality decides how many conversations you get at all: Cognism's 2026 Competitiveness Gap research found teams pairing verified contact data with AI-assisted research reached a 13.3 percent answered rate.

If the structure convinces you, the next question is usually cost. We priced a human dialler against an AI agent, line by line, in AI callers vs hiring a telemarketer in Singapore, and the use cases and pricing sections show where an agent fits and what it costs. To hear this structure delivered on a live call, book a 20-minute demo.

Frequently asked questions

What is the best cold call opening line according to data?

In Gong's analysis of 90,380 cold calls (2018 dataset), opening with the question how have you been achieved a 10.01 percent success rate against a 1.5 percent baseline, a 6.6x lift. Asking whether it is a bad time performed 40 percent worse than baseline, so drop that line entirely.

How long should a cold call last?

The average cold call runs 82 seconds (Cognism, 2026), while Gong's 2021 dataset found successful cold calls last nearly twice as long as unsuccessful ones. Treat length as an outcome, not a target: a familiar opener, a clear reason and good questions are what earn you the extra minutes.

How many questions should a sales script include?

Gong's analysis of 519,000 discovery calls (2017 dataset) found that 11 to 14 targeted questions correlates with the most successful calls, and that returns fall away beyond 14. Spread them through the conversation. On a short first cold call, three or four well-chosen questions from that bank is realistic.

When is the best time of day to make cold calls?

Cognism's 2026 cold calling report found the 10am to 11am window performs best, against an average success rate of 2.7 percent across all calls. Timing helps at the margin. The bigger levers are the opener, a clearly stated reason for calling, and consistent daily volume.

Can an AI calling agent follow a cold call script?

Yes, and it follows the script identically on every dial, so the opener, reason and questions never drop out under pressure. Every call is recorded and transcribed, which means you can review real conversations against the intended structure, spot the lines that fail, and tune the script from evidence.