The 90-Day Agency Playbook: Building ¥2M/Month in New Revenue With AI Chatbot Reselling

A Note Before the Math

Numbers in this article are illustrative models based on 2026 Japan market benchmarks, not a specific agency case study. Results depend on your client base size, sales execution, and timing.

The Simple Math

Assume your agency has 50 active clients on web or marketing retainers. You upsell AI chatbot to 20 of them at ¥15,000/month/client license. You also sell a ¥150,000 setup service to 5 new clients per month.

Result: ¥300,000 recurring + ¥750,000 project = ¥1,050,000/month from just 20 active clients.

This is a benchmark model, not a guarantee. But the math shows that with an existing client base, ¥2M/month is not an unrealistic target. This playbook is your step-by-step roadmap to pressure-test it with your own sales team.

The Math: What ¥2M/Month Actually Looks Like

Model assumptions (illustrative): Agency of 20–50 staff, client base of 50–80 Japanese SMEs, enrolled as OneBot Tier 1 partner.

Month 4 Revenue Breakdown (post-90 days)

¹ Tier 1: 70% agency / 30% VAON revenue share. Infrastructure cost ≈ ¥1,500–3,000/tenant/month.

² Setup fee is an agency service — not shared with VAON.

Note: To reach ¥2M, the agency needs an additional ~¥500K via upsell combinations (more clients, higher managed service tiers, or referral deals). At Tier 1.5 (10+ clients), pricing shifts to a flat ¥5,000/client/month structure, and the agency retains 100% of the upside above that.

Growth Trajectory Across 3 Phases

This is a trajectory, not a guarantee. Actual figures depend on your team's close rate.

The 90-Day Playbook: 3 Phases × 30 Days

PHASE 1 — Days 1–30: Foundation

Goal: Complete internal infrastructure setup, secure team buy-in, identify 3 pilot clients.

Phase 1 Objectives

• Sign partner agreement with OneBot / VAON
• At least 2–3 salespeople able to deliver a 5-minute chatbot pitch
• Demo account running on live data
• 3 pilot clients contacted and showing genuine interest

Week-by-Week Actions

Week 1: Partner Onboarding

• Sign OEM/reseller agreement with VAON (free for Tier 1)
• Set up AWS account + Gemini API key with VAON guidance (VAON-assisted)
• Receive brand kit (logo placeholder, color guide) to create white-label branding
• Build demo account with a sample knowledge base (use your own agency FAQ documents)
• Book a 2-hour kickoff call with the VAON team to walk through the roadmap

Week 2: Internal Training

• Run a 2-hour internal training session — VAON can join remotely
• Every salesperson demos the chatbot independently in under 10 minutes
• Build a simple ROI calculator (Excel/Sheets) — see the Templates section
• Prepare a 10-slide pitch deck — see the Templates section
• Rehearse the 3 most common objection-handling scenarios — see Sales Script section

Week 3: Client Selection

• Filter your client base by: (1) high volume of repetitive questions, (2) already has LINE OA or wants one, (3) currently complaining about CS costs or response times
• Shortlist 5–8 candidates; select the 3 most likely to convert as pilots
• Reach out to these 3 clients via personal email or LINE (not a cold pitch — you already have the relationship)
• Framing: "We're testing a new technology and want to try it with a few select clients before a wider launch. Would you be interested in being among the first?"

Week 4: Soft Pitch and Finalize

• Meet with the 3 pilot clients — run a 20-minute demo
• No pressure on revenue; goal is "try it at no cost / discounted price"
• Finalize pilot terms: 30 days free, or 50% discount for the first 60 days
• Kick off setup for the first client if verbal agreement is in place

Phase 1 Deliverables

• Live demo account
• Pitch deck v1
• ROI calculator
• 3 pilot clients with verbal agreement

Phase 1 KPIs

• 3 pilot clients confirmed (hard gate before moving to Phase 2)
• 100% of sales team can demo the product
• Internal NPS on the product ≥ 7/10

Common Phase 1 Pitfalls

Pitfall 1: Choosing the wrong pilot clients. Don't start with your most demanding clients to prove a point. Start with your easiest ones — you need a success story at this stage, not a battle story.

Pitfall 2: Training that stays theoretical. If your salespeople haven't deployed a chatbot hands-on at least once, they won't pitch with confidence. Hands-on demo is mandatory.

Pitfall 3: Overcommitting to clients. The pilot phase is for learning. Don't promise "AI will answer 100% of questions" when the knowledge base hasn't been fully trained yet.

PHASE 2 — Days 31–60: Pilot

Goal: Deploy 3 pilot clients, capture real metrics, build a case study draft, refine your pitch.

Phase 2 Objectives

• All 3 pilot clients running live
• At least 1 set of real numbers (questions handled, response time, client feedback)
• Pitch deck v2 with embedded case study
• 2–3 paying clients signed

Week-by-Week Actions

Weeks 5–6: Deploy and Monitor

• VAON team deploys chatbot for each pilot client (approximately 2 weeks/client, can run in parallel)
• Train the knowledge base with client documents: FAQs, product catalogs, policy docs
• Test at least 50 real questions before go-live
• Set up weekly 30-minute check-in with each pilot client
• Install dashboard tracking: conversation volume, % handled automatically, handoff rate

Week 7: Capture Metrics

• After 2 weeks live: export data from the OneBot dashboard
• Calculate: "The chatbot handled X% of questions" — typically 40–70% in the first 2 weeks depending on the knowledge base
• Ask clients: by how much has the CS team's time on repetitive questions decreased?
• Capture a testimonial quote (text or 1–2 minute video) — will be used in your pitch
• Identify gaps: which questions is the chatbot not answering well → update the knowledge base

Week 8: Refine and Upsell

• Update pitch deck with real numbers from the pilot (replace placeholders with actual data)
• Build a 1-page case study draft: Problem → Solution → Results
• Pitch converting pilot clients to paid plans
• Begin outreach to 5–8 Tier 2 clients from your shortlist (warm leads, existing relationship)
• First paying client = milestone. Celebrate internally — team morale matters.

Phase 2 Deliverables

• 3 pilot reports (1 page each)
• Case study draft v1
• Pitch deck v2 with real data
• 2–3 paying clients signed

Phase 2 KPIs

• At least 1 pilot client achieves ≥40% automation rate
• 2 paying deals closed
• 1 testimonial quote captured
• Pitch deck updated with real numbers

Common Phase 2 Pitfalls

Pitfall 1: Setting automation rate expectations too high. 40–60% is strong for the first month. The knowledge base needs ongoing updates. Don't report "AI handles 95% of questions" before you're there.

Pitfall 2: Not tracking metrics from day 1. If you don't have a baseline before deployment, you can't prove improvement. Ask clients: "How many repetitive questions does your CS team receive per day?" before go-live.

Pitfall 3: Overlooking the Handoff feature. OneBot includes a Handoff function — escalating to a human agent when the AI isn't confident. This is a key selling point for clients worried about AI missing edge cases. Don't skip it in demos.

PHASE 3 — Days 61–90: Scale

Goal: Pitch your full client base (15–20 targets), launch outbound, close 8–10 deals.

Phase 3 Objectives

• Full pitch rollout to all qualified clients in base
• Outbound campaign running in parallel (3–5 new logos from outside your current client base)
• 8–10 paying deals active by end of month 3
• Internal process standardized (onboarding checklist, support escalation path)

Week-by-Week Actions

Weeks 9–10: Full Pitch Rollout

• Email broadcast to all clients: "OneBot AI Chatbot — Results from our pilot" (attach 1-page case study)
• Book 1-on-1 meetings with top 15–20 qualified accounts (by revenue tier and industry fit)
• Use pitch deck v2 with case study — pitch results, not theory
• Offer "first-month setup waived" for clients who sign before date X (time-limited urgency)

Week 11: Outbound Push

• LinkedIn outreach to Sales/Marketing Directors at Japanese SMEs not currently in your client base
• Partner referrals: ask 2–3 partner businesses (accounting firms, HR software vendors) for cross-referrals
• Publish 2–3 LinkedIn posts on case study results (data-led, not advertising)
• Prepare a proposal template that can be customized in 30 minutes

Week 12: Close and Standardize

• Follow up on all leads from weeks 9–11 (many B2B deals close on the 2nd or 3rd follow-up)
• Finalize onboarding checklist so scaling doesn't depend on a single person
• Document the support escalation path: chatbot issue → agency point of contact → VAON support
• Internal retrospective: what worked, what didn't, plan for month 4

Phase 3 Deliverables

• 8–10 paying clients (target, not a guarantee)
• Standardized onboarding process
• Outbound playbook v1
• Monthly revenue dashboard

Phase 3 KPIs

• 8 paying clients (minimum viable to justify effort ROI)
• Close rate ≥ 40% on warm leads from existing client base
• Average deal size ≥ ¥15,000 license + ¥100,000 setup
• 1 referral deal sourced from outside current client base

Common Phase 3 Pitfalls

Pitfall 1: Pitching too many clients at once without deployment bandwidth. VAON can deploy multiple clients in parallel, but the agency needs enough capacity to onboard and support them. Don't close 10 deals simultaneously if only one person is handling it.

Pitfall 2: Skipping follow-up. In Japanese B2B, decision cycles are long. "We need more time to consider" does not mean rejection. A polite follow-up 7–10 days later is expected and necessary.

Pitfall 3: Not standardizing process early enough. If all chatbot knowledge lives in one person's head, the agency cannot scale. Write checklists, write SOPs — rough drafts are fine.

Tools and Templates: 7 Assets You Need

1. Pitch Deck (10 slides)

What it needs: Problem (CS costs in Japan), Solution (chatbot overview), How it works (RAG demo), Case study (pilot results), Pricing options, Next steps.

Note: Slide 1 is not "About our company" — it must be the problem the client is experiencing right now. The case study slide must contain real numbers, not generic testimonials.

2. ROI Calculator (Google Sheets)

What it needs: Inputs: questions/day, CS headcount, average salary. Output: current CS cost/month vs. chatbot cost + estimated savings.

Note: Don't overclaim. Let clients enter their own numbers — self-completed calculators build more trust than pre-filled ones.

3. Demo Script (15 minutes)

What it needs: Open with a real use case relevant to that specific client. Demo 3 scenarios: common FAQ, a more complex question (RAG in action), handoff when the AI doesn't know.

Note: Demo with the client's industry-specific knowledge base, not a generic one. Takes 1–2 hours to prep but meaningfully improves close rate.

4. Proposal Template (3 pages)

What it needs: Executive summary (1 page), Scope and deliverables (1 page), Pricing and timeline (1 page).

Note: Japanese proposals require formal language. Don't use machine-translated English templates. If you don't have a Japanese copywriter yet, use VAON's Tier 1 proposal draft as a base.

5. Onboarding Checklist (agency internal)

What it needs: Checklist from "contract signed" to "chatbot live": AWS setup, branding, knowledge base collection, testing, training, go-live sign-off.

Note: VAON provides a checklist template. Customize it with agency-specific steps (brand review, client training session).

6. Monthly Report Template (client-facing)

What it needs: Total conversations, % handled by AI, handoff rate, top unanswered questions, recommended knowledge base updates.

Note: This is your monthly touchpoint with clients. Strong numbers → upsell opportunity. Weak numbers → proactive support before clients complain.

7. Case Study Template (1-page PDF)

What it needs: Client profile (anonymized if needed), Challenge, Solution (chatbot features used), Results (3 specific numbers), Quote.

Note: One page, not five. Must work as an email attachment or a printout you leave in a meeting.

Pricing Strategy: 3 Models for Agencies

Model 1: License Markup (Recurring Revenue)

Structure: Agency buys from VAON at Tier 1 (70% revenue share) and resells with markup.

¹ Calculation: ¥15,000 × 70% = ¥10,500 retained by agency, minus infra ~¥2,000–3,000 = ≈¥7,500–8,500 net/client.

Best for: Agencies building a recurring revenue baseline. Easy to forecast, easy to pitch.

Note: ¥15,000 is the minimum recommended price. Many Japanese agencies currently charge ¥20,000–30,000 for chatbot services.

Model 2: Setup Fee + Lighter Monthly

Structure: One-time setup charge (agency service, not shared with VAON) plus lower ongoing license.

Best for: Clients sensitive to long-term commitment. Setup fee creates serious commitment from both sides.

Model 3: Managed Service Monthly (Full Package)

Structure: Bundle license + amortized setup + ongoing maintenance into a single higher monthly fee.

Best for: Clients without bandwidth to manage it themselves. Highest agency margin but requires capacity.

Note: If your agency isn't staffed for it, don't sell Premium to everyone. Start with Basic or Standard.

Sales Script: Objection Handling

Objection 1: "¥15,000/month is expensive — unclear ROI"

"That's a fair question. The more important question is: how much time does your CS team currently spend on repetitive questions each day? If it's 3 hours a day, that's roughly ¥30,000–50,000 in labor costs per month. OneBot handles 40–70% of those questions — starting from month one. ¥15,000 versus ¥30,000+ — the math is fairly clear. And we can run a 30-day pilot first so you see real numbers before committing."

Objection 2: "Our clients don't have an IT team"

"That's exactly why we chose OneBot — no IT team required. Setup is handled entirely by us. The client only needs to provide their FAQ documents and LINE OA access. All technical work happens on the backend. Your clients only use the product; they don't maintain it. We've already deployed this for [a similar reference client] — not a tech company — in two weeks."

Objection 3: "Our clients already have a chatbot"

"Is the chatbot they're using rule-based or AI? If it's rule-based, they're probably running into situations where questions outside the script go unanswered — and get escalated to a human immediately, which frustrates customers. OneBot uses RAG — it learns from the client's actual documents, can answer questions that aren't in any preset script, and when it genuinely doesn't know, it hands off at the right moment. I can arrange a side-by-side demo if that would help."

Objection 4: "We need more time to review internally — let's revisit later"

"Completely understood. To support your internal review, I'll send over a 2-page document: a case study from our pilot, plus an ROI calculator where you can plug in your own numbers. Ten minutes of reading should give your team the data they need for the internal discussion. I'll send it over today — does that work?"

5 Mistakes to Avoid

1. Selling chatbot as a tech product, not a business solution.

Japanese clients don't buy "RAG AI technology." They buy "reduce 60% of repetitive inquiries so the CS team can focus on higher-value interactions." Always lead with business outcomes, not features.

2. Not defining knowledge base ownership up front.

Who is responsible for updating documents when a client adds new products? If this is ambiguous from the start, three months later the chatbot will give wrong answers and the client will blame the agency. Define KB update SLAs in the contract on day one.

3. Over-promising automation rates.

Some vendors claim "AI handles 90% of questions." In practice, month-one automation is typically 40–60%, improving as the knowledge base is trained. If you promise 90% and deliver 50%, the client is disappointed — even though 50% is genuinely good. Under-promise, over-deliver.

4. Skipping the LINE OA integration.

For the Japanese market, LINE is a more important channel than a website widget for many SMEs. If a client already has a LINE OA — and most Japanese SMEs do — integrate immediately. This is a key differentiator for OneBot versus chatbots without native LINE support. See the complete LINE chatbot guide for setup details.

5. Not tracking and reporting monthly.

Client signs in month 1 → no reports → month 3 they ask "what is the chatbot actually doing?" → churn. Monthly reports are not a courtesy — they are a retention tool. Strong numbers = upsell opportunity. Weak numbers = proactive fix before the client notices.

FAQ

Q: What does the agency need to invest before generating revenue?

A: With Tier 1, there is no upfront fee from VAON. The agency's real investment is: internal training time (estimated 20–30 hours total), materials preparation (pitch deck, templates), and the time cost of running 3 pilot clients at free or discounted pricing. AWS and Gemini API costs during the pilot period can be minimal — consult VAON about the cost-sharing structure during this phase.

Q: If a client cancels after 3 months, what does the agency lose?

A: Any setup fee already collected stays with the agency. License revenue stops. Knowledge base data belongs to the client under the BYOK model. The main risk is the time invested in onboarding. Mitigate by: (1) signing 6–12 month contracts with a cancellation clause, and (2) ensuring pilot results are strong before converting to paid.

Q: How well does OneBot handle Japanese?

A: The base LLM is Gemini 2.5, which has strong Japanese-language capability. The knowledge base is trained from the client's own documents — Japanese input produces Japanese output. VAON has deployed with Japanese clients (GMJ, AiboLink). Quality depends on the quality of input documents — clearer documents produce more accurate chatbot responses. Learn more about the underlying technology in our RAG chatbot guide.

Q: How long before the agency sees ROI?

A: With the setup fee model (¥150,000) plus license (¥15,000/month), the agency breaks even on effort cost after 2–4 clients depending on overhead. License recurring revenue starts compounding from month 2 onward. This 90-day playbook is designed to have 8–10 paying clients by the end of month 3 — approximately ¥120,000–150,000 in license MRR plus project revenue.

Q: Can the agency sell managed services on top?

A: Yes. VAON provides the platform and support. The agency defines its own managed service packages (monthly KB updates, reporting, strategy sessions). This is high-margin revenue because the agency already knows the product and doesn't share it with VAON. Pricing Model 3 in the section above covers this approach. For OEM white-label structuring, see our AI chatbot OEM guide.

Next Step: Apply for the Partner Program

This playbook is the roadmap. The next step is to start.

The OneBot Partner Program (Tier 1) has no initial fee — low risk, clear upside. Ninety days from today, you could have 8–10 active chatbot clients and a new recurring revenue stream that doesn't require additional headcount.

Apply for the Partner Program →