AI PROMPTS FOR CALLS
AI prompts that make every call smarter
See how our customers use AI prompts to analyze real calls, uncover key insights, and turn conversations into data you can sort, score, qualify, and act on.
Make every field smarter with AI
Six jobs teams hand to AI
Every prompt in the library does one of these.
Pick the job, steal the prompt, change the labels to yours.
Our prompt library
Company names anonymized by industry.
Prompts lightly cleaned for clarity. Copy any of them.
JUMP TO A SPECIFIC PROMPT CATEGORY:
Tag & Categorize
Tags are how a call gets sorted the moment it ends: lead or not, which bucket, what outcome. The best tag prompts read like a decision tree, and spend most of their words defining what should NOT be tagged.
Auto-tag every call by topic
The simplest possible tagging prompt, and the most-used one across all accounts.
Determine what the call was about and tag it accordingly using the set of tags I’ve created.
Nimbata field: Tags · Most-used prompt
Why it works. Proof you do not need to be a prompt engineer to start. Point the AI at the tag set you already built and let it sort.
Storage lead vs. tenant vs. moving-equipment
Separates a real storage lead from the lookalikes (existing tenants, billing issues, U-Haul/truck rentals) that flood a facility line.
Analyze the transcript and return ONLY ONE label: “Storage Lead | Call Answered”, “Storage Lead | Call Went to Voicemail”, or “No Tag”.
1) Call type: a real human conversation is Answered; only a caller message or voicemail prompts is Voicemail.
2) It IS a storage lead if the caller asks about renting storage / parking / RV / boat / vehicle space, availability, pricing, sizes, or move-in timing, or already rents but wants more space.
It is NOT a storage lead if the caller asks about a current unit, bill, payment, or gate/account issue; is a wrong number, spam, job inquiry, or vendor; or asks about truck / U-Haul / trailer / tow-dolly rentals or moving equipment (that is transportation, not storage).
3) Voicemail is strict: only tag it if the message explicitly mentions storage / parking / vehicle space or clear intent (“looking to rent”, “need a unit”). Do not tag a vague “call me back”.
Priority: existing-tenant issue is No Tag; only rentals or moving equipment is No Tag; clear storage intent is a Storage Lead. When uncertain, return No Tag. Return exactly one label.
Nimbata field: Tags · Recommended
Why it works. Most calls to a storage facility are not leads. This prompt names the impostors and treats voicemail strictly.
Stacking disposition and outcome tags
Applies several tags at once, but keeps the mutually exclusive ones (spam, wrong number) cleanly isolated.
Reps qualify inbound callers and try to warm-transfer them live to a local partner agent. Apply exactly ONE Warm-Transfer Attempt tag.
Successful WT: a live transfer completed. The rep hands off (“I have [name] on the line…”) AND a new voice actually comes on and speaks. Giving out a phone number is NOT a completed transfer.
Unsuccessful WT: rep found a specific agent by name and tried to connect, but they were unavailable and a callback was arranged (“she’s assisting another client”, “she’ll reach out shortly”).
No PA Available: rep confirmed no partner covers that area or community (“we don’t have a partner there”, “only available in [other city]”).
Declined WT: caller is a confirmed buyer or renter but explicitly declined to be connected (“not ready to talk to an agent”).
Not Qualified: caller is not an active buyer or renter (wrong number, resident with a non-purchase question, vendor, realtor, or purchase already complete). Do NOT use if the caller expresses purchase intent anywhere in the call.
No Contact: no real conversation (IVR overflow, dropped, voicemail only).
No WT Attempt: caller was a confirmed buyer or renter but the rep never tried to find or connect an agent (just gave info or a number). The distinction from Unsuccessful WT: no agent search was attempted at all.
Nimbata field: Tags · Most-used prompt
Why it works. Proof you do not need to be a prompt engineer to start. Point the AI at the tag set you already built and let it sort.
One winning tag in a regulated intake
Picks exactly one intent tag from a defined list where only the highest value prospect should win.
Apply exactly one tag.
Qualified Lead: wants to obtain or change marketplace coverage (enroll, compare plans, get a quote, switch, or reactivate a cancelled plan).
Cancel Plan: wants to cancel, terminate, or stop paying for their current policy.
1095-A Request: asking about the 1095-A tax form or IRS reporting for marketplace insurance.
Employer Coverage: already covered through their job, not switching.
Medicaid / Medicare: has, or is asking about, Medicaid or Medicare.
Already Covered: has insurance and does not want to change.
Disconnected / Hang-up: no meaningful conversation.
Rule: only tag Qualified Lead for a genuine prospect looking to obtain or change coverage. If the call is about cancellations, tax forms, employer insurance, Medicaid, Medicare, or being already covered, use that tag instead.
Nimbata field: Tags
Why it works. A compliance-heavy line gets many inquiries. The closing rule protects the primary lead tag so tax-form or Medicaid calls never inflate lead counts.
Classify the warm-transfer outcome
Sorts every call into one of seven warm-transfer outcomes based strictly on spoken transfer signals.
Apply ALL tags that describe what happened on the call. Tags stack.
Disposition tags (apply at most ONE, or none for a normal conversation):
SPAM: solicitor, robocall, or anyone selling to the agency.
Wrong Number: caller wanted a different business or is not seeking insurance.
No Answer: not answered, went to voicemail, or hung up before connecting.
Outcome tags (apply any that are true, including several):
Current Customer: caller has an active policy, even if calling about a new quote or vehicle.
Quote: caller requested or received an auto-insurance quote.
Policy: caller bound a new policy on the call (paid and/or confirmed coverage).
Rules: if a disposition tag applies, do NOT apply outcome tags. Outcome tags stack. A sold policy is usually Quote + Policy; an existing customer quoting another vehicle is Current Customer + Quote, plus Policy if they bind.
Nimbata field: Tags · Recommended
Why it works. Leans on concrete spoken signals to draw a fine line between “tried and failed” vs. “never attempted”, crucial for referral commission reconciliation.
Tag a conversion only on real commitment
Marks “Conversion” only when the caller actually agrees and commits to dispatch on the phone.
Goal: tag a call “Conversion” only when the caller clearly commits to the service in this interaction.
Mark Conversion if the caller:
explicitly accepts (“Yes, send it”, “Book me now”, “Let’s proceed”),
gives the details to proceed now (pickup + vehicle + timing) and agrees to dispatch, or
is a returning caller confirming a prior quote (“calling back to go ahead”).
Do NOT mark Conversion for:
non-committal or deferral (“I’ll call back”, “just checking prices”),
questions only, with no agreement,
cancellations or refusals,
wrong number, spam, sales pitches, or lead brokers,
dropped or low-quality calls where commitment cannot be confirmed,
a third party who will not confirm on the owner’s behalf.
If intent is ambiguous, default to no conversion. If it is not a conversion, you may add “Inquiry” or “Quote” instead.
Nimbata field: Tags
Why it works. Prevents tire-kickers and price checkers from falsely triggering ad conversions. When in doubt, defaults to no conversion.
Score Quality & Intent
A rating turns a call into a single comparable number.
The trick is deciding what you are scoring: these prompts score commercial value and intent, not whether the agent was simply polite.
Rate commercial interest, one to five
A clean, one-line rating rubric any team can drop into Nimbata on day one.
Rate the call by its commercial interest: 1 star for a wrong number or a clearly uninterested caller, up to 5 stars for someone who sounds like a likely client.
Nimbata field: AI Rating · Recommended
Why it works. Short and calibrated. Anchoring 1 (wrong number/spam) and 5 (ready-to-buy client) gives the AI an intuitive scale.
Score the value of the call, not the service
Scores each call’s commercial intent from 1 to 5, booking potential, and category across Greek and English.
Score the COMMERCIAL value of each inbound call, not the agent’s customer service. Calls may be in Greek or English; judge meaning, not wording.
5: Very high value. Clear rental intent, close to booking; asks about or gives details like dates, pick-up/drop-off, car category, price, availability, or how to complete the reservation. A 5 does not require a completed booking.
4: Good potential. Relevant and interested, but less immediate or still comparing options.
3: Relevant, unclear value. Car-rental related but weak or unclear intent, no clear next step.
2: Low value. Weakly related, little booking intent.
1: No value. Wrong number, spam, internal or supplier call, unrelated.
Do not reward politeness or penalize a call that simply did not book. If strong intent was lost to poor handling, note it.
Also return:
Booking potential: High / Medium / Low / None
Category: New Booking, Price/Availability, Existing Booking, Insurance/Terms, Payment, Complaint, General Info, Wrong Number, Internal/Partner
Reason, and Recommended action (Follow up / No follow up / Review manually).
Nimbata field: AI Rating
Why it works. Reminds the AI not to reward politeness or penalize a call that simply didn’t finish checkout yet.
Score by job size and service area
Rates the enquiry by the scope of work and whether the caller is within the geographical service territory.
Rate 5 if the caller asks about larger tree-surgery work (tree removal, pollarding, forest management), is within our service area, AND requests a quote, site visit, or follow-up.
Rate 4 for the same larger work in-area but with no mention of a quote, visit, or follow-up.
Rate 3 for smaller jobs (pruning, stump grinding, hedge trimming) in-area.
Rate lower if the enquiry is unrelated, outside the area, or shows low intent to book.
Nimbata field: AI Rating
Why it works. Folds in both job size (tree removal vs. minor trimming) and location, so out-of-area calls never look like hot leads in reports.
A rating that gates an automation
Scores lead quality on a 1 / 3 / 5 scale and enforces a downstream rule for CRM synchronization.
Rate each call on lead quality and conversion intent.
5/5: customer booked, confirmed a time, or clearly agreed to move forward.
3/5: real interest but no booking. Asked about treatment, pricing, results, availability, or location for about two minutes or more.
1/5: no meaningful interest. Hang-ups, wrong numbers, very short calls, quick declines, or spam.
Judge the whole transcript, not keywords. A caller who asks several relevant questions is a 3 even without booking; a caller who books is a 5.
Routing rule: only send calls rated 3/5 or higher to the CRM. Do not send anything below 3/5.
Nimbata field: AI Rating
Why it works. The rating is not just passive analytics; its gating rule controls whether the call payload gets sent to HubSpot/Salesforce/Pipedrive/Zoho.
Estimate Revenue Value
A call value puts a revenue figure on each conversation to measure true CAC and feed conversion values to Google/Meta Ads.
Ranges beat guesses, and speech repair keeps numbers accurate.
Value by service type, only if booked
Assigns flat trade values ($2,000 for HVAC/Plumbing, $1,000 for Electrical) only when the job is actually scheduled.
Calculate the call value by service type. HVAC: $2,000. Plumbing: $2,000. Electrical: $1,000. Apply the value only if the job is booked.
Nimbata field: AI Call Value
Why it works. The “only if booked” clause stops the AI from attributing phantom revenue to tire-kickers, keeping ROAS truthful.
A readable value map for a service menu
Assigns clean revenue values in Euros based on which marketing retainer or project the client asks for.
Estimate business revenue from the service discussed:
Social media management: EUR 500
Google or Meta ads management: EUR 400
Website development: EUR 5,000
Funnels, lead magnets, growth hacking: EUR 1,000
Support requests: EUR 50
Existing clients asking general questions: EUR 0
Nimbata field: AI Call Value
Why it works. Clean lookup table. Often all a digital marketing agency needs to attribute conversion revenue back to Google Search Ads.
Extract a price from a mangled transcript
Pulls the exact quoted dollar amount from a conversation, repairing STT transcription errors first.
Value = the dollar amount WE quoted the caller on this call for our own service. Return a number only, no symbol. Return 0 if no amount was quoted.
WHAT COUNTS: any amount our agent states for our own work (tow, hook, lockout, jump start, tire change, winch-out, storage, mileage, or an impound total), whether or not “dollars” or “$” is said. Prices often appear as a bare number: “you’re looking at 105”, “your tow bill is 300”.
REPAIR THE TRANSCRIPT FIRST (speech-to-text mangles prices):
“15750” or “157 50” becomes 157.50; “10815” becomes 108.15.
“1 05”, “one 05”, “one of five”, “a 105” become 105.
A lone small number where a price belongs is usually a truncated hundred: “that’ll be 5 dollars” for an in-town lockout becomes 105.
Sanity-check against our real price list (105 in-town service call, 108.15 on card, 115.50, 150 or 157.50 tow, 4.50 per loaded mile, 55-65 per day storage, 200 hook, about 300 police impound, 175 per hour heavy). A candidate far from all of these is probably a misread.
WHICH NUMBER when several are said:
Cash vs card: take CASH (“105 cash, 108.15 card” becomes 105).
Conditional on the vehicle: the price for the condition that applies; if never settled, the LOWER.
A range: the LOW end (“380 to 400” becomes 380).
Split payment: ADD the parts (“90 card + 15 cash” becomes 105).
Base plus per-mile or per-day: return the BASE (“150 to hook plus 4.50 a mile” becomes 150).
A later revised price replaces an earlier one.
DO NOT COUNT: the caller’s own cash on hand, money owed to a third party, city or impound fees we are not charging, another company’s price, a card-surcharge percentage, or a “how much?” the agent never answered (0). “No charge today” is 0.
Read the whole transcript. The price is often stated late, after the address and vehicle details.
Nimbata field: AI Call Value
Why it works. The masterclass. Teaches the AI the price book, then handles cash vs. card, split payments, price ranges, and base-plus-mileage.
Estimate a contract’s value from the call
Estimates monthly recurring or project contract value based on facility type, square footage, and scope.
Estimate the potential contract value of this lead from the transcript, using realistic commercial-cleaning, janitorial, staffing, and project pricing. Pick the range that best matches facility type, scope, urgency, and service.
Recurring janitorial / day porter: small office or retail $600-$1,500/mo; medium office, warehouse, or medical $1,500-$4,000/mo; large industrial or campus $4,000-$10,000+/mo.
Staffing / employee leasing: $3,000-$15,000+/mo by headcount, shifts, hours.
Specialized / high-compliance (data center, lab, cleanroom, aerospace): $3,500-$12,000+/mo.
One-time: post-construction $500-$4,000; deep clean $400-$2,000+; floor care $300-$1,800+; carpet $250-$1,200+; emergency $500-$5,000+.
If square footage, frequency, multiple buildings, urgency, or compliance are mentioned, increase the estimate. If vague, use the midpoint of the right category. Spam or irrelevant is 0. Return ONLY the numeric USD value.
Nimbata field: AI Call Value
Why it works. B2B service contracts span $500 to $15,000+. Giving the AI realistic industry pricing bands produces defensible pipeline values.
Value the primary opportunity by service
Maps aesthetics treatments to realistic dollar amounts, picking the primary opportunity rather than summing wildly.
Estimate the revenue value of this call from the service or services discussed, using realistic ranges for an aesthetics and wellness clinic:
Botox / neurotoxin: $300-$800 (up to $1,200+ multi-area).
Filler: $600-$1,200+ single area, $1,200-$3,000+ full-face.
Medical weight loss / GLP-1: $300-$1,200+/mo.
TRT: $150-$450+/mo.
Vitamin or wellness injections: $50-$300 per visit, or $150-$600+/mo.
Skincare or facials: $150-$600+.
If several services are discussed, value the primary opportunity, not the maximum. Non-clinical calls (vendor, spam, wrong number) are 0. Return a single numeric USD value, no explanation.
Nimbata field: AI Call Value
Why it works. Callers frequently ask about multiple treatments. This prompt guides the AI to value the focal procedure and zero out vendor calls.
Summarize & Take Notes
A good summary saves someone from listening to a 10-minute recording. A great one is structured for a specific reader (marketer, sales manager, or compliance officer) with zero hallucinated fluff.
Summarize the call
The baseline summary prompt, and the second most-used across the entire Nimbata platform.
Summarize the call and add any relevant context as comments.
Nimbata field: AI Summary
Why it works. Short and versatile. Most businesses start here, then gradually add fields once they know what their reps need.
Executive summary, translated and redacted
Generates an English executive summary from a Spanish phone call, while redacting credit cards and physical addresses.
Create an executive summary in English at the top. Capture the caller’s reason for calling, key facts, buying intent, objections, sentiment, agent performance, and call outcome. Summarize the important business information clearly. Translate the notes into English from Spanish.
Do NOT capture credit or debit card numbers or address information.
Nimbata field: AI Summary
Why it works. Solves two major needs in one prompt: automated multilingual translation for unified CRM notes, plus strict PII/PCI security redaction.
Turn a lost call into structured loss data
Generates standardized loss intelligence on unconverted inquiries to analyze why deals are slipping.
Analyze the call itself (do not rely on other AI tags). Generate structured CRM notes ONLY when it is a new car-rental booking inquiry with no reservation confirmed on the call. For every other call, output exactly: NOT APPLICABLE.
Rules: English only; one line per field; no markdown; use only what was explicitly said; write “Unknown” if unavailable; never infer; never include personal data (names, phone, email, card, IDs).
Primary non-conversion reason: choose one from a fixed list (Price too high, No availability, Category unavailable, Deposit or credit-card requirement, Insurance or excess concern, Location or time unsuitable, Customer undecided or comparing, Customer will call back, Agent failed to progress, Call disconnected, Other, Unknown). Use the reason that most directly blocked the booking; do not use “will call back” when a clearer blocker exists.
Also capture: Secondary factor; Loss ownership (Customer / Pricing-policy / Availability-ops / Agent-influenced / External-technical / Unknown); Agent conversion effort (Strong / Adequate / Weak / Not assessable); Alternative offered; Follow-up owner, action, and status; Saveability (Likely / Possibly / Unlikely); Booking details (Pickup, Drop-off, Dates, Duration, Vehicle, Price, Insurance, Deposit, Payment, each “Unknown” if not stated); a one-sentence Service quality; and a one-sentence Additional context or “None”.
Nimbata field: AI Note
Why it works. Short and versatile. Most businesses start here, then gradually add fields once they know what their reps need.
A summary written for marketing
Generates an English executive summary from a Spanish phone call, while redacting credit cards and physical addresses.
Summarize the call for a dental clinic’s marketing and sales analysis. Include the caller’s main reason for getting in touch, any specific treatments mentioned (implants, Invisalign, whitening, and so on), and how far along they are in deciding: ready to book, comparing options, or just gathering info.
Note any concerns or objections (price, availability, nervousness, confusion about services). If they mention how they found the clinic (Google, an advert, social media, a referral), include it. End with the outcome (booking made, follow-up arranged, or no action), plus any quotes that reflect the caller’s attitude, urgency, or satisfaction.
If the caller asked about price, was told the price, and still did not convert (or said they would call back), add the note: “Lead not completed post pricing.” Always state whether they are a new or existing patient.
Nimbata field: AI Summary
Why it works. Includes a high-value trigger: “Lead not completed post pricing” flag, turning soft hesitation into an automated re-engagement audience.
Facts-only bullet summary with fixed fields
Strict, factual bullet summary in French with zero fluff across 4 strict fields: Problème, Équipement, Prix, Issue.
Summarize the call using factual and precise bullet points. Each line should contain only observable facts present in the conversation: no interpretation, no assumptions, do not make anything up. Preserve important details (brand, model, equipment type, exact reason for the refusal, technical constraints). Never replace a brand with a category.
Use exactly these fields:
• Problem: specific need or issue
• Equipment: type + exact brand if mentioned
• Price: stated amount or “not mentioned”
• Issue: final outcome (appointment / quote / callback / information / refusal / off-topic)
Nimbata field: AI Note
Why it works. Forbids guessing or replacing a specific brand name with a generic category. Ideal for dispatchers and field technicians.
Follow-ups & Coaching
These prompts generate the next action instead of just describing the last one: follow-up emails, structured action plans, or specific coaching recommendations for the sales rep.
Draft the follow-up email
Drafts a polished, ready-to-send recap email strictly grounded in the conversation points and agreements.
Generate a clear, concise, professional follow-up email that reflects exactly what was discussed. Include:
Brief recap: only what was actually covered (the caller’s needs, concerns, goals).
Next steps: for each, what happens next, who is responsible, the expected timeline, and any documents or approvals needed.
Tone: match the caller’s tone (only if positive) to keep rapport; keep it human and helpful, not robotic.
Optional closing offer of help or clarification.
Output the email body only.
Nimbata field: AI Note
Why it works. Adopted widely because it saves reps 5 minutes per call. Accurately recaps topics, specifies next steps, and mirrors client rapport.
Turn the call into an action plan
Extracts a prioritized task checklist with task descriptions, owners, urgency, and deadlines.
Create an action plan that strictly reflects the caller’s actual needs, concerns, and goals. For each action item use:
Description: what needs to be done
Owner: who is or could be responsible
Priority or Urgency: Low / Medium / High
Deadline: any due date or time-sensitive dependency
Follow-up requirements: documents, data, approvals, additional calls
Dependencies or blockers: anything that must happen first
Communication strategy: mirror the caller’s tone if positive
Keep it structured, easy to read, and tied directly to call content. Output a numbered list with line breaks between items.
Nimbata field: AI Note
Why it works. Shifts teams from “we talked on the phone” to accountable operational execution with dependencies and timelines.
Coach the agent from the call
Generates strengths, improvement opportunities, and actionable coaching exercises from real phone handling.
Generate a coaching plan from the agent’s performance, using only what is explicitly demonstrated in the call.
Strengths observed: effective techniques, communication wins, empathy, call control, problem-solving.
Improvement opportunities: specific behaviours to refine, tied to moments in the call; do not generalize beyond it.
Actionable coaching steps: for each, what to adjust or practice, why it matters, and how to implement it.
Recommended coaching style: how a team lead should deliver this (collaborative, supportive, direct) based on the call’s tone.
Omit any point that does not apply. Format as a list with line breaks between items.
Nimbata field: AI Note
Why it works. Turns every recorded call into a constructive mini QA session for sales managers without spending hours listening manually.
Route, Extract & Flag
Custom fields capture one clean, deterministic value per call: a route, an extracted detail, or a Yes/No gate. They connect Nimbata call data into HubSpot, Salesforce, Zapier, and webhooks.
Route intake to the right practice area
Classifies the caller’s inquiry into one practice area from a fixed list.
Identify the practice area for the caller’s inquiry. Choose one: Workers’ Compensation, Personal Injury, Car Accident, Medical Malpractice, Premises Liability, Product Liability, or Construction Accident.
Nimbata field: Custom Field · Text
Why it works. A clean classify-into-a-fixed-list pattern. Swap the list for your own service lines and it routes any intake call.
Extract the caller’s company, no guessing
Extracts the exact organization name from voicemails, with an anti-hallucination guard.
Extract the name of the company the caller says they are from in the voicemail. Return only the company name. If none is mentioned, leave empty. Do not guess or infer from the phone number or email.
Nimbata field: AI Note
Why it works. Enforces a strict anti-hallucination constraint: no guessing from email domains or caller ID. Produces spotless CRM account records.
Should this call go to the CRM?
A binary YES/NO decision gate that filters out support, spam, and vendor noise from the sales pipeline.
Should this call be added to the sales CRM as a potential opportunity?
Answer “YES” only if the call is a new or potential sales opportunity.
Answer “NO” if the caller is an existing customer with no additional-sale opportunity, the call is purely support, admin, or informational, or there is no realistic potential for future business.
Nimbata field: Custom Field · Yes/No
Why it works. Binary fields are ideal for webhook triggers in Zapier, Make, and HubSpot. Keeps your sales pipeline 100% focused on real opportunities.
Map the need to a service dropdown
Extracts the exact organization name from voicemails, with an anti-hallucination guard.
Choose the correct service based on the prospect’s needs from: PVC/Aluminum Windows, Roller Shutters, Sliding Glass Doors, Entry Doors, Garage Doors, Pergolas, Gates, Awnings, Fences, Carports, Heat Pumps, Insulation, Solar Panels, Heating, Ventilation, Calculate My Grants, Other.
Nimbata field: Custom Field · List
Why it works. Matches CRM picklist options character-for-character so data seamlessly routes to specialized trade estimators.
AI prompts for calls: Your questions, answered
AI prompts for calls are plain-English instructions you give an AI model to extract something specific from a phone call. Instead of listening back to a recording, you write an instruction like “perform sentiment analysis on the call and apply the right tag” and the AI reads the transcript and fills the field for you. Applied across every call, they turn recordings into structured, filterable, reportable data.
Call scoring is rating each phone call against a consistent set of criteria so you can compare leads objectively. Traditionally a manager listens to a sample of calls and fills in a scorecard by hand. With AI prompts you score every call automatically, on buying intent, lead quality, urgency, or your own qualification checklist, which means your cost per qualified lead reflects reality rather than a 5% sample.
Yes. Once a call is recorded and transcribed, AI can summarize it, detect sentiment, extract the caller’s exact words, score intent and assign a revenue value, all without anyone listening back. In Nimbata you write the instruction once and it runs on every inbound call from then on.
Yes. The prompts read text, not audio, so each call needs to be transcribed first. Nimbata records and transcribes automatically, so the transcript is there before you write your first prompt. If you are using these prompts elsewhere, you will need to export the transcript and paste it in.
There are two ways, depending on whether a price actually gets said out loud on your calls.
Either catch the value mentioned on the call. If your team quotes prices, the number is already in the transcript. E.g. “Extract the value quoted on this call.”
→ £180 · “that’ll be £180 for the callout and diagnostic”
Or adapt it to your own budget model. If nobody says a price, calculate it from whatever your business actually runs on.
Real estate: “Value the lead at 1.5% of the budget the caller mentioned.” → $6,300 from a $420,000 budget
Insurance: “Value the lead at 12% of the annual premium discussed.” → $186 on a $1,550 policy