Answers to common questions about WhatsApp automation, AI agents, ad campaigns, conversion tracking, broadcast marketing, costs, and working with us.
The audit is a 30–45 minute consultation where we review your current setup - whether that's WhatsApp, CRM, ads, or manual processes - and identify the biggest opportunities for automation and AI.
You'll walk away with a clear picture of where you're losing leads, what can be automated, and a recommended system architecture tailored to your business. If it makes sense to work together, we'll outline the scope and timeline. If not, you still keep the insights.
There's no charge and no obligation. We do this because the businesses that see the gaps are the ones that move forward.
Most deployments go live within 3 to 4 weeks, following our structured five-phase process: kickoff and discovery (Week 1), system architecture and AI agent design (Week 2), build and internal testing (Week 3), soft launch with live traffic (Week 4), and full deployment with ongoing optimisation (Week 5 onwards).
The exact timeline depends on complexity. A single-agent setup with basic automations can launch in as little as 2 weeks. A multi-location enterprise deployment with custom API integrations, multiple AI agents, and complex routing rules might take 5 to 6 weeks.
We set clear expectations during the kickoff call and keep you updated at every milestone.
After we launch your system, you have two options.
First, you can run independently. We train your team thoroughly during the build process, including hands-on sessions with documentation, so you're fully capable of managing the system yourself.
Second, you can add our monthly retainer (from $799/month), which gives you ongoing AI agent optimisation, WhatsApp broadcast campaign management, monthly performance reviews, new automation builds as your business evolves, knowledge source updates for AI accuracy, and priority support.
Most of our clients choose the retainer because the system performs best when it's continuously refined based on real conversation data. AI agents get smarter over time, but only if someone is actively tuning them.
This is actually how many of our clients come to us. They've invested time and money into a GoHighLevel or respond.io setup, maybe followed some YouTube tutorials or hired a freelancer, but it's not performing the way they expected.
The free audit is designed exactly for this situation. We'll review your current configuration, identify what's working, what's misconfigured, and what's missing entirely. Then we give you a prioritised action plan.
Sometimes the fix is straightforward, like a broken automation or a poorly designed AI prompt. Other times, we recommend a more thorough rebuild. Either way, you'll know exactly where you stand and what it takes to get to where you want to be.
Yes, but we handle the entire setup for you. The WhatsApp Business API is different from the regular WhatsApp Business app - it's what enables AI agents, automations, broadcasts, and multi-user team inboxes.
As part of our onboarding, we'll guide you through Meta Business verification, set up your WhatsApp Business API account through either respond.io or GoHighLevel, and configure your phone number. This typically takes 3–5 business days.
If you're already using the WhatsApp Business app, we'll migrate your number to the API so you keep your existing chat history and contacts.
We work with any service-based or appointment-based business that handles customer enquiries on WhatsApp. Our current clients span automotive detailing, home renovation and furnishing, fitness and wellness studios, dance academies, e-commerce, property, financial advisory, insurance, F&B, beauty salons, and short-term rental hospitality.
The common thread is businesses that deal with inbound enquiries, need faster response times, and want to automate repetitive conversations like pricing questions, bookings, follow-ups, and review collection.
If you're handling 20+ WhatsApp enquiries a day and your team is struggling to keep up, the system will work for you regardless of industry.
Every AI agent we build is trained on your specific knowledge base - your services, pricing, policies, and FAQs. It doesn't make things up or pull from the internet. If the AI doesn't know the answer, it's designed to say so and escalate to your team rather than guess.
We also set guardrails on what the AI can and can't do. For example, it won't quote prices it hasn't been given, won't make promises outside your policies, and won't handle sensitive topics without human handover.
During the first 2–4 weeks after launch, we actively monitor conversations and fine-tune responses. Accuracy improves with every interaction because we're reviewing real data and adjusting the knowledge base. Most of our agents hit 90%+ accuracy within the first month.
Yes. Our AI agents detect the customer's language automatically and respond in kind. This works across English, Mandarin, Malay, Tamil, Bahasa Indonesia, and many other languages spoken by your customers.
The AI handles language-switching mid-conversation seamlessly - no manual language routing needed. We've deployed multilingual agents for clients in automotive, home services, F&B, and e-commerce, all operating in mixed-language environments daily.
You can, at any time. Every system we build includes a shared team inbox where your staff can see all conversations in real time. If someone wants to step in, they simply take over and the AI stops responding for that conversation.
You can also set rules for automatic handover - for example, if a customer asks for a refund, mentions a complaint, or requests to speak to a person. The AI will hand off gracefully and notify your team immediately.
Think of it as AI handling the volume while your team handles the exceptions. Most of our clients find that 80–90% of conversations are fully resolved by the AI without any human involvement.
No. If you run ads, this system makes them dramatically more effective. Most businesses lose 60–80% of ad-generated leads to slow replies and lack of follow-up. AI agents reply instantly, qualify the lead, and book the appointment before they lose interest. Same ad spend, more conversions.
But if you're not running ads, you likely have an existing customer database across spreadsheets, CRMs, POS systems, and booking records. That database is one of your most valuable assets. WhatsApp broadcasts get 70–90%+ open rates compared to around 20% for email. When customers reply, AI agents pick up the conversation naturally. Re-engage past customers, promote new services, drive repeat visits - all without your team manually chasing anyone.
Whether your growth comes from paid traffic or your existing network, the system pays for itself.
It comes down to what your business already has in place and what you're trying to achieve.
If you don't have an existing CRM and want a single platform that handles your pipeline, funnels, automations, email, SMS, and WhatsApp all in one place, GoHighLevel is usually the right fit. It's ideal for appointment-based businesses, sales teams running multi-step funnels, and companies that want one login for everything.
On the other hand, if you already have a CRM like Salesforce or HubSpot and your primary need is AI-powered conversations layered on top of your existing stack, respond.io is typically the better choice. It excels at high-volume WhatsApp operations, multi-agent inboxes, and deep CRM enrichment.
We make this recommendation during the free audit after understanding your tech stack, team size, and goals. There's no pressure either way because we're certified on both and our pricing is identical regardless of platform.
Yes, and this is actually one of the most common scenarios we handle. We integrate with Salesforce, HubSpot, Zoho, GoHighLevel, Pipedrive, and custom APIs.
During setup, we map your existing fields, lifecycle stages, and data structures so nothing gets lost or duplicated in the migration. If you're using a CRM we haven't listed, we can usually connect it through API or webhook integrations.
We also make sure your team's existing workflows aren't disrupted. The goal is to enhance what you already have, not force you to start over on a new platform.
Yes. We run Click-to-WhatsApp ad campaigns on Meta (Facebook and Instagram) that feed directly into the conversation systems we build.
This means a prospect sees your ad, taps to message you on WhatsApp, and immediately gets greeted by an AI agent that qualifies them and books an appointment - all within the same conversation. It's a closed loop from ad spend to booked revenue.
We handle campaign strategy, creative direction, audience targeting, and budget optimisation - all connected to your conversation infrastructure so you can track exactly which ads are driving real bookings and revenue, not just clicks.
We run four core campaign types across Meta (Facebook and Instagram), Google, and TikTok:
Traffic campaigns drive volume to your WhatsApp number or landing page. Best for awareness and building your contact list quickly.
Conversion campaigns are optimised for a specific outcome - a completed booking, a qualified lead, or a purchase. The ad platform's algorithm learns from real conversion data (sent back via server-side tracking) and progressively targets higher-intent audiences.
Lead generation campaigns capture contact details through in-platform forms or direct WhatsApp conversations, then feed them into your CRM and automation sequences for instant AI-powered follow-up.
Click-to-WhatsApp (CTWA) campaigns on Meta are our specialty. Instead of sending people to a website, the ad opens a WhatsApp conversation directly. An AI agent greets the prospect, qualifies them, and books the appointment - all within one chat. CTWA leads get a 72-hour free messaging window from Meta, making them significantly more cost-effective than traditional lead forms.
On Google, we run Search campaigns (for high-intent keyword traffic) and Display campaigns (for retargeting and brand awareness). On TikTok, we run traffic, conversion, and lead generation campaigns optimised for the platform's high-engagement audience.
During the free audit, we recommend the right mix based on your goals, budget, and where your audience spends time.
When someone taps a Click-to-WhatsApp ad on Facebook or Instagram, it opens a WhatsApp conversation with your business number instead of sending them to a landing page. Here's what happens next:
Your AI agent greets the prospect within seconds, using context from the ad (the campaign, ad set, and creative that triggered the click are all tracked). The agent qualifies the lead by asking relevant questions - budget, timeline, service needed - and either books an appointment, provides a quote, or routes to your sales team for complex cases.
Because the conversation happens inside WhatsApp, you capture the contact's phone number automatically. There are no forms to fill out, no landing page drop-off, and no fake email addresses. Meta also provides a 72-hour free messaging window for CTWA leads, so your follow-up messages within that period cost nothing.
The AI agent tags and scores each lead in your CRM based on their responses, so your team always knows which leads are hot and which need nurturing. If a lead goes quiet, automated follow-up sequences re-engage them at timed intervals.
This closed-loop system - ad click to WhatsApp conversation to qualified booking - is why CTWA campaigns consistently outperform traditional lead form campaigns on cost-per-qualified-lead.
Read our full tracking architecture guide →
Yes. We configure Meta Conversions API (CAPI) and TikTok Lower Funnel Events so your ad platforms learn from real conversions - qualified leads, completed purchases, booked appointments - not just clicks.
This is server-side tracking, so it works regardless of browser privacy settings or cookie restrictions. Businesses using CAPI typically see up to 60% fewer spam leads within weeks.
Traditional pixel-based tracking relies on a JavaScript snippet in the user's browser to fire events back to Meta. The problem is that browser privacy features, ad blockers, iOS restrictions, and third-party cookie deprecation all prevent the pixel from firing reliably. Meta gets incomplete data about which ad clicks actually convert, and its algorithm optimises based on partial information.
Meta Conversions API (CAPI) solves this by sending conversion events server-to-server. When a real business outcome happens - a lead is qualified, an appointment is booked, a purchase is completed - your conversation platform sends that event directly to Meta's servers along with the original CTWA click ID that was automatically captured when the prospect clicked the ad.
Because this happens at the server level, it is completely unaffected by browser privacy settings, ad blockers, or cookie restrictions. Meta receives a clean, complete signal of what is actually working.
The practical impact is significant. Meta's ad algorithm needs a minimum of roughly 50 conversion signals before it can meaningfully optimise delivery. With pixel-only tracking losing a large portion of events, it can take weeks to accumulate enough data. With CAPI, events are captured reliably from day one, so the algorithm learns faster, targets better audiences, and wastes less of your ad budget on low-quality leads.
We configure CAPI as part of every ad management engagement. The supported event types include Lead Generated and Purchase, and the events must occur within 7 days of the ad click to count toward optimisation - which is another reason our AI agents are configured to qualify and convert leads quickly rather than letting them sit in a pipeline.
TikTok Lower Funnel Events is TikTok's equivalent of Meta's CAPI - it allows us to send real conversion events (such as qualified leads, completed purchases, and booked appointments) from your server directly to TikTok's ad platform, bypassing browser-based tracking limitations entirely.
When someone interacts with your TikTok ad and enters a WhatsApp conversation or completes a form, we capture the attribution data. As the lead progresses through qualification, booking, or purchase, the relevant events are sent back to TikTok server-side. TikTok's algorithm then uses these signals to find more people who are likely to take the same high-value actions, rather than optimising for superficial metrics like clicks or video views.
This is especially valuable on TikTok, where engagement patterns differ from Meta. By feeding TikTok real business outcomes, we help the algorithm move beyond vanity metrics and deliver leads that actually convert.
We set this up alongside Meta CAPI for clients running cross-platform campaigns, so both ad platforms learn from the same quality signals.
Businesses using CAPI typically see measurably better ad performance within the first few weeks. The most commonly reported improvement is up to 60% fewer spam or low-quality leads, because Meta's algorithm is learning from genuine conversions rather than noisy browser signals. Over time, cost-per-qualified-lead drops as the algorithm gets better at targeting high-intent audiences.
The key requirement is volume: Meta needs approximately 50 conversion events before its optimisation kicks in meaningfully. This is why we configure AI agents to qualify leads quickly and send conversion events promptly - every event must be sent within 7 days of the original ad click to count toward optimisation.
There are geographic restrictions to be aware of. CAPI is not currently available for businesses targeting audiences in the EU/EEA, the United Kingdom, or Japan due to ePrivacy regulations in those regions. If your campaigns target those markets, we'll discuss alternative tracking and optimisation strategies during the audit.
For all other markets, CAPI works globally and we configure it as standard practice for every ad management engagement.
WhatsApp messaging is charged per conversation window, not per message, and Meta bills you directly. When a customer messages you first, the entire 24-hour conversation is free - this is typically 70–90% of your traffic.
You only pay when your business initiates a conversation outside an active window using a pre-approved template. Marketing broadcasts are the most expensive category, followed by utility messages (reminders, confirmations) and authentication messages (OTP codes). Click-to-WhatsApp ad leads get an extended 72-hour free messaging window, making them especially cost-effective.
Rates vary by country - Meta sets pricing differently for each market. Use our cost calculator to estimate your monthly costs based on your specific market and messaging volume.
Zelix does not mark up Meta or respond.io fees.
WhatsApp Business API charges per template message delivered, not per individual text. There are three paid message categories: Marketing (promotions, broadcasts), Utility (transactional updates like reminders and shipping notifications), and Authentication (OTP codes, login verification).
When a customer messages your business first, a 24-hour service window opens. All replies within that window are free, including free-form messages and utility templates. You only pay when you send a template message to someone outside an active window.
Click-to-WhatsApp ad leads get an extended 72-hour free messaging window, making them especially cost-effective for businesses running paid traffic to WhatsApp.
Meta sets WhatsApp Business API rates differently for each country or region based on local market conditions, purchasing power, and WhatsApp adoption levels. Countries with higher business messaging demand tend to have higher rates, while high-volume markets have lower per-message costs.
Some countries do not have individual pricing and fall under regional rates. Rates are set by Meta and apply equally regardless of which platform provider you use (respond.io, WATI, Twilio, etc.).
Yes. When a customer sends the first message, a 24-hour service window opens. During this window, your business can send unlimited free-form messages, images, documents, and even utility template messages at no charge from Meta. The window resets with each new customer message.
This means most customer support conversations cost nothing. The only time you pay is when your business initiates a conversation outside an active window using a pre-approved template message (e.g., sending a broadcast, reminder, or OTP to someone who hasn't messaged in the last 24 hours).
The most effective way to reduce costs is to segment your broadcast lists. Instead of blasting your entire contact database, target only contacts who are likely to engage. Sending to 1,000 engaged contacts costs half of sending to 2,000 where half are inactive.
Run Click-to-WhatsApp ads instead of traditional lead forms. Leads from these ads get a 72-hour free messaging window, meaning your first interaction is free.
Use AI agents to qualify and respond within the free window so you rarely need to initiate paid conversations. Send utility messages (reminders, confirmations) promptly after a customer interaction while the free window is still open - if the customer messaged you today, the reminder you send tonight is free. If you wait until tomorrow, it costs a utility fee.
No. Meta bills WhatsApp conversation fees directly to the business through the WhatsApp Business API account holder. Neither Zelix Labs nor respond.io add any markup to Meta's messaging rates.
respond.io charges a separate platform subscription fee for the software (AI agents, workflows, broadcast tools, CRM features). Zelix Labs charges separately for setup, automation configuration, and ongoing optimisation. All three costs are independent and transparent.
The Annual Broadcast Marketing Plan is a full-year broadcast strategy and execution retainer. Instead of sending ad-hoc messages whenever you remember, we build a comprehensive 12-month playbook that maps every broadcast campaign to your business calendar - seasonal promotions, product launches, holidays, re-engagement windows, and quiet periods.
The retainer includes: strategic planning (annual calendar with monthly broadcast themes), content creation (message copy, creative assets, and template approval with Meta), audience segmentation (targeting the right contacts for each campaign to maximise open rates and minimise costs), execution (scheduling and sending through respond.io or GoHighLevel), and performance reporting (open rates, reply rates, conversion tracking, and cost analysis after each campaign).
This service is designed for businesses that know broadcast marketing works but lack the time or expertise to plan, write, and execute consistently. Instead of sporadic one-off blasts, you get a structured programme that builds momentum over the year.
The plan is reviewed and adjusted quarterly based on what's performing, so it stays responsive to real results rather than following a rigid script.
We start with a strategy session where we map your business calendar: peak seasons, slow periods, product or service launches, holidays relevant to your market, and any recurring events (anniversaries, membership renewals, etc.).
From there, we build a 12-month broadcast calendar with a recommended cadence - typically 2 to 4 broadcasts per month, depending on your audience size and industry. Each broadcast is categorised by purpose: promotional (driving sales or bookings), educational (tips, guides, updates that build trust), re-engagement (targeting dormant contacts), and transactional (reminders, confirmations, follow-ups).
We segment your contact list for each broadcast so that the right message reaches the right audience. Sending a fitness promotion to your entire database wastes money and annoys contacts who aren't interested. Sending it only to contacts who've previously enquired about fitness services costs less and converts better.
Each quarter, we review performance data - open rates, reply rates, conversion rates, and messaging costs - and adjust the plan for the next quarter. Campaigns that performed well get expanded; underperformers get reworked or replaced.
The result is a broadcast programme that feels intentional to your customers rather than random, while keeping your WhatsApp messaging costs predictable and controlled.
No. The initial build is a one-time project fee - you pay, we build, you own the system. There's no ongoing commitment tied to the build itself.
The monthly retainer is optional and runs month-to-month. You can pause or cancel anytime with 30 days' notice. We don't lock clients in because we'd rather keep you through results than through a contract.
That said, the platform subscriptions (respond.io or GoHighLevel) are billed directly by those platforms on their own terms, which are typically monthly or annual. We'll walk you through the options during onboarding.
Yes. We offer flexible payment options including milestone-based payments split across the project. This is common for our Scale and Enterprise plans where the build spans several weeks.
We accept credit card payments (processed securely via Airwallex) and direct bank transfer. We'll work out a payment schedule that fits your cash flow during onboarding - no awkward conversations needed.
Absolutely, and we take this seriously because several of our clients operate in regulated industries like financial advisory, insurance, and healthcare.
All systems we deploy use end-to-end encryption for conversations in transit and at rest. We implement role-based access controls so only authorised team members can see customer data. Our setups are compliant with applicable data protection regulations including PDPA and GDPR, with proper consent management, data retention policies, and audit trails built in.
Critically, no customer conversation data is ever used to train AI models. Your business data stays your business data, full stop.
We're happy to walk through our security practices in detail during the audit if data handling is a key concern for your business.
You do. All customer data - conversations, contact records, CRM entries, broadcast lists - belongs to your business. Zelix Labs builds and configures the system, but we do not own, resell, or retain any of your customer data.
Your WhatsApp Business API account is registered to your company, your CRM instance belongs to you, and your platform subscription (respond.io, GoHighLevel) is in your name. If you ever stop working with us, your data stays exactly where it is. We don't hold anything hostage.
During the engagement, we access your systems with credentials you provide in order to build and optimise. You can revoke our access at any time by changing permissions in your platform's admin settings.
Your data is stored by the platforms your system runs on - not by Zelix. Each platform has its own infrastructure and data residency policies:
respond.io hosts data on cloud infrastructure with servers in multiple regions, and holds ISO 27001:2022 certification. GoHighLevel uses AWS infrastructure with data centres primarily in the United States. WhatsApp Business API message data is handled by Meta's infrastructure with encryption in transit and at rest.
If data residency is a compliance requirement for your business (for example, data must stay within a specific country or region), we'll flag this during the audit and recommend platforms or configurations that meet your requirements.
Nothing changes. Your WhatsApp Business API account, CRM data, conversation history, contact lists, and automation workflows all stay in your platform accounts. These are yours - they don't disappear when you stop working with us.
If you were on our retainer, we'll hand over any broadcast calendars, campaign reports, and documentation we created during the engagement. We'll also walk your team through system management so they can continue operating independently.
We recommend revoking our admin access to your platforms after offboarding, which you can do from your platform's user management settings. We don't retain copies of your data externally.
AI agents are configured with strict guardrails around sensitive topics. By default, the AI will never ask for or process payment card details, bank account numbers, or government-issued ID numbers within WhatsApp conversations. If a customer volunteers sensitive information, the AI is trained to acknowledge it without storing or repeating it, and to escalate to a human agent immediately.
For clients in regulated industries (financial advisory, insurance, healthcare), we configure additional safeguards: mandatory human handover for specific conversation topics, restricted AI responses on compliance-sensitive subjects, and audit trails that log every conversation for regulatory review.
Payment processing is always handled through secure, external payment links or gateways - never through the chat interface. Medical or financial data that needs to be recorded is routed to your secure CRM with appropriate access controls, not left in chat transcripts.
WhatsApp requires businesses to obtain opt-in consent before sending template messages (broadcasts, reminders, promotions) to customers. Meta enforces this strictly - violating consent policies can result in messaging restrictions or account suspension.
We build consent management into every system from day one. This includes: opt-in collection through your website, landing pages, or in-store forms; clear disclosure of what the customer is subscribing to; easy opt-out mechanisms in every broadcast message (typically a "Reply STOP to unsubscribe" flow); and automated list hygiene that removes opted-out contacts immediately.
For broadcasts specifically, we maintain segmented lists where only opted-in contacts receive marketing messages. Transactional messages (appointment reminders, order confirmations) have different consent requirements - customers who initiate a transaction are generally considered opted in for related transactional updates.
We document your consent flows during setup so you have a clear audit trail if questions arise.
Only the team members directly assigned to your project. We operate on a least-privilege basis - no one at Zelix gets blanket access to all client systems.
During the build phase, our project lead and technical team need admin access to configure your platform (respond.io, GoHighLevel, CRM). Once the build is complete and you're running independently, you can remove our access entirely.
If you're on the retainer, we maintain access for ongoing optimisation, but this is limited to the specific platforms we manage for you. We never access systems or data outside the scope of our engagement. You can audit who has access at any time through your platform's user management panel, and we encourage clients to review access permissions periodically.
No. There's an important distinction here: your AI agent is trained on your knowledge base (your services, pricing, policies, FAQs), but customer conversations are never used to train the underlying AI model itself.
The AI models we use (through respond.io and other platforms) process conversations in real time to generate responses, but the conversation content is not fed back into model training. Your customers' messages remain private and are not used to improve AI for other businesses.
What we do with conversation data is review it for quality assurance - identifying where the AI gave incorrect or suboptimal responses so we can update the knowledge base and improve accuracy. This is a manual, human-reviewed process, not automated model training. The purpose is to make your specific agent better, not to train a general AI model.
15-minute call. Action plan within 2 working days. No sales pressure.
WhatsApp Us