If you use (or will use) Kommo CRM with WhatsApp, the fastest way to stop “thinking it’s expensive” and start deciding with data is to calculate the real cost per useful conversation and the cost per sale. It’s not just the Kommo license: it includes WhatsApp API, number, provider/BSP, messages by category, automation, integrations and mainly human time. In this article I give you a practical model (with formula and table) so you can close this calculation and unlock ROI without breaking costs when scaling.
I am Luiz Otávio Gonçalves (mechanical engineer turned Digital Engineer) and I implement Kommo and GoHighLevel in Brazil focusing on execution and ROI. There are no miracles here: just metrics, process, and optimization.
The biggest mistake that destroys ROI on WhatsApp (and why you feel like “the numbers don’t add up”)
The number 1 mistake is adding only “CRM monthly fee” + “a WhatsApp provider” and ignoring the rest. Result: you scale volume, customer service becomes a bottleneck, response rate drops, leads cool off, and the team asks for more people. Then WhatsApp “gets expensive” — but in practice the expensive part is the process without instrumentation.
If you want a complete implementation map (to avoid bans, reduce cost per service, and increase conversion), I go deeper into this in: Kommo CRM with WhatsApp API: implementation checklist that prevents bans, reduces cost per service, and unlocks ROI.
Who this calculation is for (and who it is NOT for)
- Serves for operations that sell via WhatsApp with volume (inbound, outbound, referrals), have more than 1 seller/agent and want Predictability to control costs when scaling.
- Serves for clinics, real estate, B2B industry, services, and agencies that depend on follow-up and have sales cycles of days/weeks.
- DOES NOT serve if you just want “a nice WhatsApp” without funnel discipline and without measuring conversion. If you won’t measure, any number becomes an excuse.
- DOES NOT serve if your ticket is so low and your margin so tight that you have no room to pay for human service: then the path is e-commerce/self-service or reducing WhatsApp scope.
What makes up the real cost of selling via WhatsApp in Kommo
I divide it into 5 blocks. The idea is for you to be able to put this into a spreadsheet in 30 minutes and stop deciding in the dark.
- 1) CRM: Kommo licenses per user (sales team + customer service + management).
- 2) WhatsApp Channel (official API): number, provider/BSP and charges per conversation/message (depends on Meta’s current model).
- 3) Automation: bot, routing, fields, rules, integrations (n8n, webhooks, etc.).
- 4) Team: salary + charges + the invisible cost: time spent on repetitive tasks and rework.
- 5) Losses due to process: leads without response, slow response, lack of follow-up, duplicates, absence of history. This is a cost because it lowers conversion.
Want a mature way to measure the return on all this? I detail the method here: How to measure the ROI of business automations.
The indicator I use: cost per useful conversation (CPU) + cost per sale (CPV)
You don’t want “cost per message.” You want cost per useful conversation (conversations that become real opportunities in the funnel) and cost per sale.
Practical definition (no philosophy):
- Useful conversation = conversation that generates measurable progress: minimum qualification, scheduling, proposal sending, negotiation, closing or post-sale.
- Sale = closed/won (obvious) or equivalent event (e.g., enrollment, signed contract, invoiced order).
Formulas (simple, to run every month)
1) Total Monthly Channel Cost (CMTC)
CMTC = Kommo (licenses) + WhatsApp API (conversations/messages + number/provider) + Automation/integrations + Cost of the team dedicated to WhatsApp
2) Cost per Useful Conversation (CPU)
CPU = CMTC / Number of useful conversations in the month
3) Cost per Sale (CPV)
CPV = CMTC / Number of sales attributed to WhatsApp in the month
4) ROI (direct model)
ROI = (Gross margin generated via WhatsApp − CMTC) / CMTC
Without margin, ROI becomes fiction. If you don’t know margin, start with the basics: average ticket and approximate margin. Then you refine.
Real cost table: model for you to fill in (without guessing numbers)
I won’t guess WhatsApp billing values here because Meta operates with rules and categories that change and vary by country/structure (and “made-up number” is the fastest way for me to make you make a wrong decision). What I will do is give you a table ready for you to fill in with what you pay today.
| Component | What goes in | How to measure | Where money usually leaks |
|---|---|---|---|
| Kommo CRM | Licenses per user + add-ons if any | Kommo monthly invoice | “Ghost” user and team without CRM routine |
| WhatsApp API | Usage (conversation/message by category) + number + BSP | BSP invoice + usage report | Poorly used template, unnecessarily reopened conversation, lack of screening |
| Automation | n8n/integrations, webhooks, servers, maintenance | Subscriptions + internal/third-party hours | Automation without owner, broken flow, manual rework |
| Team | Agents/SDRs/sellers (portion of time on WhatsApp) | (Cost/hour) x (hours on the channel) | Answering everything manually, without macros, routing, or priority |
| Losses due to process | Lead without response, high SLA, weak follow-up | % without first response + average time + win rate | “Seller’s WhatsApp” without history and without governance |
The real data you need to collect (in 7 days) to close the calculation
If you want to calculate CPU and CPV without mistakes, you need a minimum of instrumentation. I recommend collecting this for 7 days and projecting for the month:
- Volume: how many active conversations per day and by source (ads, organic, referral, outbound).
- time: average time for first response (SLA) and total time spent per conversation.
- Quality: % of conversations that reach the qualification stage (useful conversation).
- Conversion: win rate by source and by agent.
- Cost: invoices (Kommo + BSP + tools) and team cost/hour.
If your operation is still setting up commercial automation the right way (process before tool), this guide helps avoid hacks: How to create automation flows from scratch.
How to reduce cost per conversation without hurting conversion (what I implement in practice)
Now the important part: optimization. Reducing cost is not “answering less.” It is answering better, faster, and with less human touch where it’s not needed.
1) Automatic screening (before reaching a human)
You unlock ROI when the human only steps in where there is a real chance to advance. Good screening does three things:
- Identifies intention (budget, scheduling, support, billing, order status).
- Collects minimum data (name, city, need, deadline, investment range).
- Routes to the right queue (sales vs support) and right person (region, product, seniority).
This reduces time per conversation and increases useful conversations. When you measure, CPU drops without you “strangling” the channel.
2) Aggressive SLA + automatic follow-up (without becoming spam)
WhatsApp is speed. If your SLA exceeds 5–10 minutes during business hours (in competitive niches), you are paying to generate leads for your competitor.
- activate automatic first response reminder.
- implement follow-up cadence at funnel stages.
- optimize messages by context (not “good morning” every day).
Want a cadence reference that works without annoying the lead? I already wrote about the classic mistake here: Automatic sales follow-up: what it is, why almost everyone does it wrong, and how to set it up.
3) Standardize “macros” and response library (with governance)
Without standardization, each agent reinvents the wheel. This wastes time and lowers conversion. What I implement:
- Library of responses by stage (opening, qualification, objections, scheduling, proposal, post-sale).
- Fields in the CRM to personalize without typing everything.
- Review monthly: which macros convert more (DATA > guesswork).
4) Tag and measure: where profitable conversations come from
WhatsApp itself is not “expensive” or “cheap.” Expensive is conversation without margin. If you don’t tag origin and assign sales, you finance a bad channel.
- activate tracking by origin (UTM, form, landing page, ad, list).
- to build simple reports: useful conversations by origin, CPV by origin.
- Cut campaign that generates volume but no margin.
5) When Kommo is WORTH MORE (and when it’s WORTH LESS) on WhatsApp
My implementer’s view:
- Kommo is worth more when you want to centralize WhatsApp + Instagram/DM + funnel with automation and service governance, without relying on the salesperson’s personal WhatsApp.
- Kommo is worth less when your operation is extremely simple (few leads, low volume, almost automatic sale). Then it can be overkill and you pay implementation curve without return.
If you’re unsure whether Kommo is overkill for your scenario, I went straight to the point here: Is Kommo worth it for small businesses and freelancers? When does it pay off and when is it overkill?.
Quick checklist: what I validate before scaling WhatsApp (to avoid exploding costs)
- Funnel with stages that represent real progress (not “conversation 1, conversation 2”).
- SLA monitored and clear goal per shift.
- Screening active automation (at least to separate sales vs support).
- Distribution of leads with rules (round robin, priority, capacity).
- Cadence follow-up with limits and logic (no spam).
- Report weekly: useful conversations, proposals, closings, CPV.
What usually changes the ROI game (in practice): reduce “human time per sale”
I’ll be very direct: the biggest ROI lever is not “paying less for CRM.” It is reducing human time per sale while maintaining (or increasing) conversion. When you reduce 10 minutes per conversation in 1,000 conversations/month, you just bought time — without hiring anyone.
And here Digital Engineering comes in: I don’t improvise. I design process, instrument data, and only then automate. The tool is the means, not the end.
Conclusion: real cost per useful conversation is the KPI that gives you decision power
When you calculate CPU and CPV, you stop arguing “opinion” and start discussing “leverage.” You can choose a plan, adjust automation, calibrate the team, and scale predictably.
If you want me to design this with you (funnel, official WhatsApp API, automation, routing, reports, and governance) and keep it running with an ROI goal, the next step is simple: request a project.