The most repeated number in sales training is a ratio. Talk 43 percent of the time, listen 57. Gong published it in 2016 after running the numbers on recorded calls, and it has been quoted in sales advice ever since.
Gong went back and re-ran the analysis on 326,000 sales calls that lasted at least ten minutes. The ratio did not survive as a dividing line. Across all calls, the average seller talks 60 percent of the time. On deals that closed, sellers talked 57 percent. On deals that were lost, 62 percent. Five points separate winning from losing, and Gong’s own write-up says plainly that “that difference isn’t massive” (gong.io).
Now read a different line in the same analysis. High performers hold roughly the same talk ratio whether a deal ends in a win or a loss. Low performers swing ten points, from 54 percent talk time on deals they won to 64 percent on the ones they lost.
That is the number worth building on, and it is not about volume at all. It is about spread. The weaker sellers are not reliably too talkative. They are inconsistent, and the version of them that shows up on a bad Tuesday is a measurably different salesperson than the one who closed on Thursday. Which means the useful question is not how much you talk. It is how far you drift.
A script is a variance control, not a persuasion device
Almost all script advice is written as though the problem were the words. Find a better opener, a sharper value line, a cleverer way to ask for the close. That framing assumes your bad calls are bad because you said the wrong thing.
The data points somewhere less flattering and more fixable. Your bad calls are bad because you did something different. You were rushed, or the prospect sounded skeptical in the first ten seconds, or it was the fourth call of a day that started badly, and you started filling silence. The script’s job is to hold the parts of the conversation that should not move, so that the parts that should move have room.
That reframing has a practical consequence. It tells you how much to write down. A script that covers the whole call will be abandoned the first time a prospect says something unexpected, which is every call. A script that fixes only the high-variance moments survives contact.
Fix two things on the page and leave the rest alone
The first thing to fix is the opening, because it is the part you perform rather than discuss. Everything after the first twenty seconds is a response to a live human being. The first twenty seconds is a monologue you deliver identically to everyone, which makes it the one stretch of the call where consistency is entirely within your control and drift is entirely your own doing.
The second thing to fix is the question set, and here the same analysis carries a genuinely counterintuitive finding. Sellers who won asked 15 to 16 questions per call. Sellers who lost asked more, about 20 (gong.io). More questions correlated with worse outcomes.
The obvious reading is that too many questions feels like an interrogation, and Gong makes that point. There is a second reading worth sitting with, because it changes what you do about it. A question count is not really a behavior. It is a symptom. Somebody asking twenty questions is usually someone who has run out of plan and is buying time, or who has not decided in advance what they actually need to know and is therefore asking everything in the hope that something lands.
So do not treat 15 as a target to hit. Treat it as a ceiling that forces a decision. Write down the questions you need answered before you can tell whether this person is worth a proposal, cap the list, and put them in an order. The count then takes care of itself, and the nervous improvising has nowhere to live, because the next question is already on the page.
Leave the middle unscripted. When a prospect raises a real objection, you are no longer delivering anything; you are sorting what kind of no you just heard, which is a separate skill and one we took apart in sorting the four kinds of no.
If you sell to consumers by phone, part of your script is already written
This is the section most sales script advice skips, and it is the one with legal consequences attached.
The Federal Trade Commission’s Telemarketing Sales Rule requires that on an outbound call a telemarketer “promptly disclose” certain information, where promptly means before any sales pitch is given: the identity of the seller, that the purpose of the call is to sell goods or services, and the nature of the goods or services being offered (ftc.gov). Read that as a script instruction and it is oddly helpful. Your first three sentences have a required payload, and the fastest way to satisfy it is also the fastest way to stop sounding like a cold caller pretending not to be one.
Two more rules shape who you are allowed to call at all. Sellers and telemarketers must delete numbers on the National Do Not Call Registry from their calling lists at least every 31 days, and the FTC lists civil penalties of up to $53,088 for each violation (ftc.gov). That is per violation, not per campaign, which is why this belongs in a small business’s process and not in its good intentions.
The exemption most small businesses actually live inside is the established business relationship. The FTC describes two kinds. One runs from a customer’s purchase, rental, lease or financial transaction and lasts 18 months from the date of the last payment, transaction or shipment. The other runs from a consumer’s inquiry or application about your goods or services, and “exists for three months starting from the date the consumer makes the inquiry or application” (ftc.gov).
Translate that into the work. The homeowner who asked you for a quote in August is callable through November even if their number sits on the registry. The customer whose last invoice cleared eighteen months ago is not, as of that date. Most small businesses have a pile of quote requests that went cold and no idea that the pile has an expiry date on it. That is a calendar reminder, not a strategy.
If you sell to other businesses, most of this lifts. The FTC lists business-to-business calls among the calls not covered by the Rule, with narrow exceptions for retail sales of nondurable office or cleaning supplies and for soliciting sales or charitable contributions from employees (ftc.gov). Worth knowing which side of that line your business sits on before you build a calling list, because the two situations have different first sentences.
Build the question set out of calls you have already had
Here is where the AI part of this is genuinely useful, and it is not the part people reach for first.
Asking a model to write you a sales script produces competent, generic copy, because the model has never met your buyers. It does not know that three of your last ten callers asked whether you are licensed before they asked about price, or that everyone from one neighborhood asks about parking. The questions your buyers actually ask are sitting in your own conversations, and until recently getting them out meant listening to recordings with a notepad.
That is the thing that changed. Transcription is now cheap enough to be a routine habit rather than a project, as we covered when the price of listening collapsed. Three real options, with prices read from each vendor’s own pricing page today.
Fathom is the most generous starting point for a one-person business. Its individual Free plan is listed at “$0” and “Free forever” with “Unlimited recordings + transcriptions”, Premium at “$16 /month / per user”, Team at “$15 /month / per user” with a two user minimum, and Business at “$25 /month / per user”, which adds “Coaching metrics & AI scorecards” (fathom.video).
Otter.ai is worth a look specifically because it reports “Keywords and speaker talk time”, which is the metric this whole piece is about. Its Basic plan is free with “300 monthly transcription minutes”, Pro is listed at “$8.33 /user/month” on annual billing, and Business at “$19.99 /user/month” (otter.ai).
Fireflies.ai is the option if you want the analysis rather than the transcript. Free is “$0”, Pro is “$10” per seat per month billed annually or “$18” monthly, and Business is “$19” per seat per month billed annually or “$29” monthly, which is the tier that adds “Conversation intelligence” and “Team analytics” (fireflies.ai).
One honest limitation, because it decides whether any of this applies to you. These tools join video meetings on Zoom, Google Meet and Microsoft Teams. If your sales conversations happen on an actual telephone, they do not capture them, and you need recording through your phone system instead.
Recording also has a consent question attached, and it is worth getting right. Federal law provides that it is not unlawful for a person “to intercept a wire, oral, or electronic communication where such person is a party to the communication or where one of the parties to the communication has given prior consent” (law.cornell.edu), so as a participant you are covered federally. Some states set a stricter standard. California, for one, makes it an offense to record a confidential communication “without the consent of all parties” (leginfo.legislature.ca.gov), so check your own state before you switch recording on. Announcing it in the first sentence costs you nothing and settles the question.
Then do the boring part. Read ten transcripts and write down every question the buyer asked you, not the ones you asked them. The recurring ones are your script’s real content, because they are the things a buyer needs settled before they can say yes. Answering the top three before they are asked is what makes a scripted opening feel like competence rather than a pitch. A model is good at this specific job, which is pattern matching over text you already own, and if you would rather start from a structured prompt than a blank page, the Sales Script Creator at BusinessPrompter.com is built for it, filed under Marketing and Sales and described as writing effective sales scripts for the different stages of the sales process. It sits behind that site’s Pro membership.
The part I would argue hardest
The Small Business Administration’s Office of Advocacy counts 36.2 million small businesses in the United States, of which 82.3 percent have no employees at all (advocacy.sba.gov). For most readers of this, there is no sales team to align. There is you, on the phone, between two jobs.
That is exactly why the consistency finding matters more for a small business than for the sales floor it was measured on. A big team’s variance averages out across thirty sellers. Yours does not average out at all. It shows up as a month that was fine followed by a month that was not, with no explanation you can name, because the difference was never in the market. It was in how the conversation went on the days you were stretched.
And there is a version of this worth being careful about. A written script is the thing that makes it possible to hand selling to a first employee without the business suddenly sounding like a different company, which is how a one-person operation becomes a two-person one. Used the other way, as a set of lines someone must recite while their talk time is monitored, it produces exactly the flat, rehearsed call everybody says they hate receiving. The difference is whether the script fixes the parts that should not vary and then trusts the person with the conversation, or tries to own the conversation itself.
What this will not do
A script will not fix an offer nobody wants. If the calls die at the point where you explain what you do, the problem is upstream of the wording, and the fix is the unglamorous work of naming what you are actually for, which is the job the value proposition canvas exists to do.
It also will not survive being read aloud. The opening is written to be said the same way every time, not recited word for word, and the difference is audible. Gong’s own suggestion for the rest of the call is smaller than a script and probably harder: pause for two seconds before responding, which stops you stepping on the end of an answer that was about to get interesting (gong.io).
And none of this requires new software. If you already pay for tools you are not using, that is worth settling before adding another subscription, which is a question we worked through in what to actually pay for.
Frequently Asked Questions
Do I need a sales script if I am the only person who sells?
Yes, and arguably more than a team does. The variance the research found is within individual sellers, not only between them: low performers’ talk time swung from 54 percent on deals they won to 64 percent on deals they lost (gong.io). A solo owner has nobody else’s good week to cancel out their bad one, so the drift shows up directly in the month’s numbers.
Will a script make me sound robotic?
It will if you script the whole call. Fix the opening, where you are performing the same twenty seconds for everyone, and fix the list of questions you need answered. Leave the middle of the conversation unwritten. The tell of a robotic call is not that the opening was prepared, it is that the seller kept delivering prepared material after the buyer said something real.
Can I call someone whose number is on the Do Not Call Registry?
Sometimes, and the windows are specific. The FTC’s established business relationship exemption covers 18 months from a customer’s last payment, transaction or shipment, and three months from the date a consumer makes an inquiry or application about your goods or services (ftc.gov). Outside those windows you need the number scrubbed, and lists have to be updated at least every 31 days.
Do I have to tell people I am recording the call?
Federal law permits recording by a party to the conversation (law.cornell.edu), but some states require consent from everyone on the call. California prohibits recording a confidential communication “without the consent of all parties” (leginfo.legislature.ca.gov), so the answer depends on your state and on where the person you called is sitting. Announcing it in the first sentence is the simple path, and in practice it rarely costs a call.
Record five calls before you write a word
The instinct when you decide your selling needs work is to sit down and write a better script from scratch. Do the opposite. Record the next five conversations, then write down only two things: what you said in the first twenty seconds each time, and every question the buyer asked you.
The first list will show you how much your opening already moves, which is the actual problem the research identified. The second list is your script, and you did not have to invent any of it.
