I have an ai voice agent startup selling minutes that our agent talks. Its been 1 year…
- User
I have an ai voice agent startup selling minutes that our agent talks. Its been 1 year and we do not have significant stable revenue? Is it normal? We did 100 poc but still no revenueWhat should we do to get more. What is the most effective sales channel for my company. We got revenue from 3 different clients that our co founder’s connections. Should we focus that or is it no sustainable? How about reaching target companies from Linkedin with connection request? 👉 For E-commerce Sites:
Calling the customer who abandoned the cart with a short voice call, reminding them and convincing them to buy by offering a special offer.
Accelerating the sales opportunity by making the “arrived” notification for the product that is back in stock with a voice call.
For Real Estate Offices:
Returning the customer who clicked on Facebook and Google ads and filled out the form on the website with a voice call instantly, getting a meeting date and preventing potential loss.
Notifying the potential customer about the listings added to favorites or whose price has decreased with a voice call.
Introducing the newly added listings to the portfolio to the customers with a voice call.
For Automotive Dealers:
Calling the person who requested a test drive with an automatic call, determining a suitable date and time and directing them to the showroom.
Making a service appointment by reminding the vehicle owners whose maintenance time is approaching with a voice call.
Making a voice call to the customer who made an offer but did not get a response, reminding them of their decision before the offer period expires. To announce special campaign and gift opportunities to loyalty program members via voice call.
👉 For Tourism and Hotel Management:
To inform the guest who makes a reservation via voice call before check-in, to offer additional service or discounted opportunities.
To provide participation information and tips to the customer who registers for a tour via voice call, to create additional service sales opportunities.
To celebrate the customer who has a birthday or anniversary via voice call and offer a special offer.
To inform potential customers about the campaign opportunity abroad via voice call.
👉 For Health and Clinic Services:
To inform the patient who has made an appointment via voice call before the appointment, to get approval and to reduce the risk of empty appointments.
To convey the renewal offer to the customer whose subscription is about to expire via voice call.
👉 For Finance and Payment Systems:
To remind the customer whose payment date is approaching via voice call and to transfer easy payment options.
👉 For Telecom Companies:
To announce the new campaign and tariff to the target customer group via voice call.
👉 For Restaurants and Cafes:
Inviting regular customers with a voice call on special menu or event days.
👉 For Educational Institutions:
Giving participation approval and program information to the person who registered for the event or seminar with a voice call.
👉 For Technology Stores and Services:
Giving the customer appointment time and transaction information with a voice call before the device arrives for service.
- Assistant
1. First ‑ breathe. One year of “interesting pilots but no steady $$” is very common for B2B infrastructure-type start-ups
• Voice/AI infra usually needs integration, security reviews, legal sign-off, etc.
• 12–24 months before repeatable revenue is almost a cliché. The key is turning PoCs into Expansion, not just stacking more PoCs.2. Why did the 100 PoCs NOT convert? — do a quick post-mortem
Possible reason Symptom you might have seen Fix “Nice demo, but ROI unclear” Stakeholders liked the tech, stalled at budget time For each vertical build an ROI calculator (show extra sales, reduced no-shows, etc.) Integration friction IT/legal/security blocking Provide ready-made Zapier/Shopify/Salesforce plugins, security docs Pricing mismatch They loved it till they saw the per-minute quote Offer tiered SaaS subscriptions (“X free minutes + seats”, overage per min) No internal champion After pilot, nobody “owned” it During pilot nominate one person whose KPI you improve Do 5–10 structured interviews with past PoC customers this week. You’ll learn more than in 100 cold LinkedIn messages.
3. Pick ONE beach-head vertical and ICP
Your use-case list is huge. Focus signals are missing. Criteria to choose:
- High value / time pressure on calls (>$ profit per conversion)
- Short security & compliance cycle
- Clear, measurable KPI you improve
- Existing systems have APIs you can plug into quickly
From experience the best early verticals for voice-AI have been:
• Automotive dealers (lead to showroom = $$$)
• Healthcare/clinics (no-show reduction = measurable)
• Real-estate brokers (commission big, decision maker is local owner)Pick one, maybe two. Build the solution, not a toolbox.
4. Design the “Land → Prove ROI → Expand → Refer” motion
- Land: ≤30-day pilot, fixed price, limited scope, clear success metric (e.g. “reduce no-shows 20%”).
- Prove ROI: weekly report e-mailed automatically.
- Expand: pre-built upsell packages (extra languages, more locations, outbound campaigns).
- Refer: ask for intro to 2 peers, give them free minutes as thank-you.
5. What sales channels work best for this kind of product?
Ranked by early-stage efficiency:
- Founder-led, relationship sales (what gave you 3 clients)
– Yes, keep doubling down. It is sustainable until $1-2 M ARR. - Channel partners / BPO call-centres / CRM integrators
– They already sell “voice” and get paid on minutes. - Targeted outbound (LinkedIn + email + phone)
– Works if you focus on one ICP and show case-study ROI. - Content / webinars / SEO around “reduce clinic no-shows”
– Good for pipeline in 6–12 months, but slower.
Cold LinkedIn alone, with no vertical focus and no hard numbers, will be low yield (<1% meeting rate). Use it as one step in a multi-touch cadence.
6. How to run LinkedIn outbound without burning time
- Define persona (e.g. “Dealer Principal, 20–200 cars / month, US/CA”).
- Build a 4-step sequence:
• Connect with short, hyper-specific note.
• Day +2 send 1-sentence value + 20-sec Loom video.
• Day +5 send mini-case study (PDF, 1 page, metrics).
• Day +9 ask: “Worth a 15-min chat or kill this thread?” - Parallel email & phone if data available.
- Use small batches (50-100 contacts / week), refine based on reply rates.
7. Partnerships – your “force multiplier”
• CRMs (HubSpot, Pipedrive)
• Call-centre platforms (Five9, Aircall, Twilio partners)
• Vertical SaaS (Dealer-management software, Practice-management for clinics)They already have the customer, billing relationship, and an “app store”. Revenue-share 20-30% is normal.
8. 90-day execution plan
Week 1–2
✔ Interview 10 past PoCs → document blockers & wins
✔ Select one beach-head vertical & “killer use case”Week 3–4
✔ Rewrite website & decks to talk only to that ICP
✔ Create ROI calculator + 2-page case study (even if from co-founder client)Week 5–8
✔ Run 30-day pilot with two new prospects, charge symbolically $1k (skin in the game)
✔ Instrument metrics for daily dashboardWeek 9–12
✔ If pilots hit KPI, sign annual contract, press-release + testimonial
✔ Launch targeted outbound + seek 1 channel partnerGoal after 90 days → 3 paying customers on annual plans, MoM MRR growth ≥20 %, one partner deal signed.
9. Key early-stage metrics to track
• Pilot-to-paid conversion rate (aim >30 %)
• Net revenue retention (expand vs. churn)
• Gross margin per minute (>60 % ideally)
• Sales cycle length (first call → closed)
• Payback on customer acquisition (<9 months)10. Mindset shift
You are no longer selling “AI magic minutes.”
You are selling “+18 % more showroom visits for car dealers” (or whichever ICP you choose).
Talk KPI, not technology. The revenue will follow.
Need a sounding board on which vertical to pick or help crafting the first outbound sequence? Just let me know and we’ll sketch it together. 🚀
Other models, same or similar prompt
Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.