Plain-English definitions of the WhatsApp Business API, AI agents, Click-to-WhatsApp ads, server-side tracking, the CRM platforms we deploy, and the compliance terms behind them. 113 terms and growing.
Who is allowed to read, write, or change which piece of customer data inside an AI deployment, and the audit trail proving those rules are being followed.
AI ImplementationThe measure of whether the team actually uses the AI deployment day-to-day, or quietly works around it. The real metric of a successful project.
AdsMeta's automated campaign type that hands more decisions (audience, placement, creative selection) to its machine-learning stack. Less control, often better results once seeded with conversions.
AIAn LLM-powered conversation agent trained on your business knowledge base, designed to answer customer enquiries, qualify leads, and escalate to humans only when needed.
AI ImplementationThe early-stage workshop where a business looks at its own operations and identifies the workflows where an AI employee would actually move the needle, before any code is written.
AI ImplementationThe two building blocks of most AI integrations: an API lets your agent ask another system for information; a webhook lets the other system push events at your agent in real time.
TrackingThe time period within which a conversion is credited back to a specific ad click or view. Different platforms use different defaults; the wrong setting can radically distort which campaigns look profitable.
WhatsAppA WhatsApp template specifically for sending one-time passwords (OTPs) and login verification codes. Tightly scoped, with strict formatting rules from Meta.
Splitting your contact database into smaller, targeted groups based on behaviour or attributes, so each broadcast reaches only the people most likely to engage.
WhatsAppA Meta-approved company that provides businesses with access to the WhatsApp Business API, plus the inbox software and tools to actually use it. respond.io and GoHighLevel are BSPs.
The total amount spent to acquire one new paying customer, usually calculated by dividing total acquisition spend (ads + sales team + tools) by new customers won in the same period.
AI ImplementationThe organised set of communication, training, and process steps that take a business from 'before AI was here' to 'AI is part of how we work' without losing people or productivity.
MetricsThe percentage of customers who stop being customers in a given period. The single most important retention metric for subscription businesses; a major input to LTV calculations.
AdsMeta ads that, when tapped, open a WhatsApp conversation with the advertiser instead of sending the user to a landing page.
AI ImplementationThe decisions about where an AI deployment runs (your tenant, a vendor's tenant, on-premise hardware), how it scales, and how it stays available when things go wrong.
AIThe maximum amount of text (measured in tokens) an LLM can consider in a single conversation. The hard ceiling on how much knowledge base, history, and current message the model can hold at once.
WhatsAppThe free messaging period that opens when a customer messages your business first (24 hours) or taps a Click-to-WhatsApp ad (72 hours). Outside this window, business-initiated messages cost a template fee.
AdsA Meta or Google ad campaign optimised against a specific conversion event (purchase, lead, booking) rather than clicks or views. The right default for any business with a measurable outcome.
MetricsThe percentage of visitors, leads, or contacts who complete a defined action: a sign-up, a purchase, a booking, a reply. The simplest funnel-health metric.
TrackingMeta's server-to-server tracking endpoint that sends real conversion events directly to Facebook, bypassing browser blockers, ad blockers, and iOS privacy restrictions.
AI ImplementationThe ongoing work of keeping the monthly bill of an AI deployment under control: picking the right model for each step, caching answers, batching calls, and watching the token meter.
AdsA list of specific people uploaded to Meta or built from your pixel and Conversions API signal. Used to retarget existing customers, suppress them from acquisition, or seed lookalike audience generation.
The work of removing duplicates, fixing typos, standardising formats, and deleting bad records from a dataset before an AI agent or model gets near it.
AI ImplementationAdding human-supplied tags to a dataset (intent, sentiment, category, outcome) so a model has examples to learn from or an AI agent has structure to route on.
AI ImplementationStandardising the way the same kind of data is represented across all your systems, so a customer name, a phone number, or a date means the same thing everywhere.
AI ImplementationThe plumbing that moves data from where it lives (your CRM, your ad platform, your billing system) to where your AI agent or model needs it, on a schedule that keeps it fresh.
ComplianceThe geographic location where customer data is physically stored. Increasingly a regulatory requirement (especially in finance, healthcare, and government) and an enterprise procurement gate.
WhatsAppThe business name that appears at the top of every WhatsApp chat with your customers. Set once during WhatsApp Business API setup, must match your registered business, and is approved by Meta.
ComplianceSingapore's Do Not Call Registry. A list of phone numbers whose owners have opted out of marketing messages. Sending marketing WhatsApps or SMS to a registered number without prior consent is a PDPA breach.
ComplianceA contract between a data controller (your business) and a data processor (a vendor handling personal data on your behalf). Required by GDPR and recommended under PDPA. Defines who is liable for what.
CRMA pre-built series of automated messages sent to a contact at fixed intervals after a trigger event. The standard way to nurture leads who are not ready to buy yet.
Storing data in an encrypted form on disk, so even if the storage media is stolen or copied, the data cannot be read without the encryption keys. The standard for any system holding customer data.
ComplianceA scheme where messages are encrypted on the sender's device and only decrypted on the recipient's device, so no intermediary (including the messaging platform) can read them in transit. WhatsApp's default for consumer chats.
AI ImplementationWhat happens when an AI workflow breaks: how it recovers, how it tells someone, and what it records so the next person to look can diagnose it quickly.
AI ImplementationThe structured process of measuring an AI agent's outputs against a known set of correct answers, so you can prove the agent is good enough before you ship and stays good enough after.
AI ImplementationThe signals that wake an AI workflow up and tell it to do something: a customer message, a payment received, a no-show on a calendar, a CRM field changing.
The ongoing work after launch of catching the 1 percent of inputs the AI agent gets wrong, understanding why, and fixing the prompt, the data, or the workflow so it does not happen again.
AIA capability that lets an LLM-powered AI agent call out to external tools (booking system, CRM, calculator, search) instead of just generating text. The bridge between the model and the real world.
MetricsThe sequence of stages a prospective customer moves through from first awareness to closed sale. Visualised as a funnel because volume drops at each stage.
Google's current web analytics product, replacing Universal Analytics. Event-based, privacy-first, cross-device, with built-in BigQuery export for serious analysis.
ComplianceThe EU's data protection regulation. Applies to any business handling personal data of EU residents, even if the business itself is outside the EU.
CRMAn all-in-one platform combining CRM, sales pipelines, funnels, email, SMS, and WhatsApp in a single tool. A good fit for businesses that want one system instead of five.
TrackingA free Google product that lets non-engineers add and update tracking scripts (Meta Pixel, GA4, conversion tags) on a website through a UI, without redeploying code.
AIRules and constraints that prevent an AI agent from going off-script: refusing certain topics, escalating sensitive cases, never quoting unverified facts. The safety layer around production AI deployments.
When an AI model generates content that sounds plausible but is factually wrong. The single biggest risk in deploying AI agents to real customers without grounding.
AIWhen an AI agent passes a conversation to a human team member. The reliability of handover is what makes AI agents safe to deploy on real customers.
CRMA mid-market CRM and marketing automation platform. Easier than Salesforce, deeper than GoHighLevel. Strong for B2B teams running content marketing, email, and sales pipelines together.
AI ImplementationAn AI workflow pattern where specific decisions are routed to a human for review or approval before action, instead of letting the agent act autonomously.
The written sequence of milestones from discovery to production, with dates, owners, and the dependency chain between each step, so an AI deployment ships on time instead of drifting.
ComplianceAn international standard for information security management. Certification means an independent auditor has verified that the organisation has the right systems, controls, and processes to protect data.
The structured set of business-specific documents and facts an AI agent draws on to answer customer questions. Accuracy and freshness of the knowledge base directly determine accuracy of the agent.
A Meta ad format where the user fills out a form inside Facebook or Instagram without leaving the platform. Cheaper per-lead than landing-page ads, but historically lower lead quality. Often beaten by Click-to-WhatsApp.
CRMA numeric score assigned to a lead based on their attributes and behaviour, used to prioritise which leads the sales team should call first.
CRMA label assigned to each contact in your CRM marking how far along they are in the customer journey: lead, marketing-qualified, sales-qualified, opportunity, customer, evangelist.
AIThe class of AI models (GPT-4, Claude, Gemini, Llama) that power modern chatbots and AI agents. Trained on internet-scale text to predict likely next words.
AdsA Meta-generated audience of new people who statistically resemble a seed audience you provide (often your best customers). The most powerful tool for finding new buyers who behave like your existing ones.
MetricsThe total revenue a single customer generates across the full duration of their relationship with your business. Often shortened to LTV; sometimes called CLV (Customer Lifetime Value).
A no-code automation platform (formerly Integromat) that connects apps and APIs through visual workflows. The middle layer in many marketing stacks where the heavy custom logic lives.
WhatsAppA pre-approved WhatsApp message used to start a marketing conversation outside an active window. Promotions, offers, broadcast campaigns. The most expensive of the three template categories.
WhatsAppThe five states a WhatsApp message moves through: sent, delivered, read, replied, failed. Visible to the sender as the familiar one-tick / two-tick / blue-tick indicators.
WhatsAppThe daily cap on how many unique users your WhatsApp Business number can initiate conversations with. Tiers are 250, 1k, 10k, 100k, and unlimited. Earned by maintaining a Green quality rating.
WhatsAppMeta's central admin console for managing Facebook pages, Instagram accounts, WhatsApp Business Accounts, ad accounts, pixels, and team access. Where every onboarding to the WhatsApp Business API actually happens.
TrackingA JavaScript snippet that fires from a webpage to tell Meta about visitor behaviour: page views, clicks, purchases. The browser-side counterpart to the Conversions API. Increasingly unreliable on its own.
AI ImplementationThe data your AI needs that nobody has been recording, or that lives in someone's head, or that exists only on paper, or that was deleted last quarter for storage reasons.
AI ImplementationComparing two or more candidate AI models on the same eval set so you can pick the right one for your use case based on data, not on hype.
AI ImplementationFurther-training a base AI model on your own data so it produces outputs that match your business style, tone, or domain knowledge out of the box, without lengthy prompts.
AI ImplementationThe instrumentation that records what your AI agent is doing in production: every input, every decision, every output, every error, with enough detail to audit, debug, and improve.
A long-running series of messages designed to build relationship and trust with a lead who is not yet ready to buy. Distinct from a drip sequence in tone: nurture is conversational and educational, drip is more transactional.
The customer's explicit, recorded permission for your business to message them on WhatsApp. Required by Meta and by most data-protection laws. The foundation of every legal broadcast list.
Singapore's data protection law, governing how businesses collect, use, disclose, and protect personal data. Comparable in spirit to the EU's GDPR, applied to Singapore-resident data subjects.
WhatsAppThe Meta process of confirming that a business legitimately controls the phone number it wants to use on the WhatsApp Business API. The first onboarding step and a common bottleneck.
CRMA position in your sales pipeline: where an opportunity sits between first contact and closed-won. Used for forecasting, sales-team prioritisation, and identifying where deals get stuck.
AI ImplementationThe small rewrites to your standard operating procedures, escalation flows, and team roles that an AI deployment forces, and the discipline of actually doing them instead of just adding the AI on top.
AIThe text input given to an LLM that shapes its response. The whole craft of working with LLMs lives in writing good prompts.
AI ImplementationA sequence of prompts where the output of one step becomes the input to the next, used when a single prompt is too complex to handle reliably.
AI ImplementationThe discipline of writing instructions for an LLM precisely enough that it gives the right answer in the right format, every time, even on edge cases.
Quality assurance (the technical team's last check) and user acceptance testing (the actual users' last check) before an AI deployment goes live.
WhatsAppMeta's traffic-light score (Green, Yellow, Red) for your WhatsApp Business phone number, calculated from how recipients react to your messages over the past 7 days. Influences how many messages you can send per day.
An AI architecture where the model retrieves relevant documents from a knowledge base before generating an answer. The standard way to ground LLM responses in real, current business facts.
CRMA messaging-first business platform built around WhatsApp, Instagram, Messenger, and SMS. Strong at high-volume conversation operations and AI-agent workflows. Zelix is a Certified Strategy Partner.
AdsShowing ads to people who have already interacted with your brand (visited the site, added to cart, opened a chat) rather than to cold audiences. Cheaper, higher-converting, but capped by audience size.
AdsRevenue generated divided by ad spend. The most common ad-performance KPI, but easy to mislead with if attribution windows or revenue recognition are not held constant.
AI ImplementationA simple cost-versus-return spreadsheet that pressure-tests an AI deployment against the actual numbers of your business before you spend a dollar building it.
The largest enterprise CRM. Powerful, customisable, and complex. The default choice for mid-market and enterprise sales teams; usually overkill for SMEs.
AI ImplementationThe state of a business where the information an AI employee needs to do its job lives in five different systems that do not talk to each other.
ComplianceThe set of controls (technical and organisational) that keep an AI deployment safe to operate under the regulations your business lives under and the trust your customers expect.
TrackingSending conversion events from your server directly to ad platforms, instead of relying on browser scripts that get blocked by privacy settings, ad blockers, and iOS restrictions.
MetricsThe percentage of booked appointments that the customer actually attends. Critical for service businesses; often where revenue leaks happen.
ComplianceAn American security audit framework focused on five trust principles: security, availability, processing integrity, confidentiality, privacy. Type II reports cover operational evidence over time.
AI ImplementationThe work of getting every person whose buy-in matters (operations, finance, legal, customer-facing team, leadership) on the same page about what the AI does and does not do, before it goes live.
AI ImplementationThe wiring that lets your AI agent read from and write to the systems your business already runs on, instead of being a standalone tool nobody connects to.
AIThe persistent set of instructions an AI agent gets at the start of every conversation: who it is, what tone to use, what rules to follow. The character sheet for the agent.
A setting on an LLM that controls how varied or predictable its responses are. Lower values produce focused, deterministic answers; higher values produce more creative, surprising ones.
WhatsAppPlaceholders inside an approved WhatsApp template (like {{1}} or {{name}}) that get replaced at send time with personalised values. The mechanism that makes a single template usable for thousands of personalised sends.
AI ImplementationThe external SaaS tools and APIs your AI deployment depends on beyond the model itself: messaging platforms, payment processors, calendars, voice transcription, search APIs, and so on.
TrackingTikTok's server-to-server conversion tracking, the equivalent of Meta's CAPI, sending real lead and purchase events directly to TikTok's ad platform.
AIThe unit LLMs use to count and process text. Roughly equivalent to a sub-word: about 4 characters or three-quarters of an English word. Costs and context-window limits are measured in tokens.
AI ImplementationThe structured way you teach your team to use a new AI deployment well: not just how to log in, but what the agent does, what it does not, and what their role looks like next to it.
WhatsAppA six-digit PIN that protects your WhatsApp Business API number from being moved to a different account or device. Mandatory on the API and recommended on the regular WhatsApp Business app.
Picking which one or two AI use cases to ship first by scoring each candidate on value, feasibility, and time-to-deploy, instead of trying to do everything at once.
WhatsAppA pre-approved WhatsApp message for transactional updates: appointment reminders, order confirmations, shipping notifications. Cheaper than marketing templates.
TrackingA query string appended to a URL (utm_source, utm_medium, utm_campaign, etc.) that lets analytics tools attribute the visitor to a specific ad, email, or campaign source.
A database that stores text as numeric vectors so an AI agent can retrieve the most semantically similar passages to a user's question, rather than matching exact keywords.
WhatsAppThe green checkmark badge next to your business name in WhatsApp, awarded by Meta to notable brands that prove their identity. Visual signal that the account is genuinely the business it claims to be.
The Meta-managed business account that owns your WhatsApp phone number, display name, message templates, and pricing tier. The container for everything WhatsApp-API-related.
CRMA real-time HTTP callback from one system to another, triggered by an event. The most common way to connect WhatsApp platforms, CRMs, and AI workflows together.
WhatsAppThe programmatic interface Meta provides for businesses to send and receive WhatsApp messages at scale, with multi-user inboxes, automation, and AI agents.
WhatsAppMeta's free mobile app for small businesses with a single phone, single user, and basic features. The simpler cousin of the WhatsApp Business API.
WhatsAppA native product catalog inside WhatsApp Business, letting customers browse items, see prices, and add to cart without leaving the chat. The foundation of WhatsApp Commerce.
WhatsAppThe end-to-end shopping flow inside WhatsApp: catalog browsing, cart, checkout, payment confirmation. Built on top of WhatsApp Catalog plus payment integrations.
WhatsAppInteractive forms that run inside a WhatsApp message, letting users complete bookings, sign-ups, or product configuration without leaving the chat. Multi-screen, branchable, native-feeling.
AI ImplementationCodifying a business process so it runs end-to-end with little or no human action: triggers, conditions, branching logic, AI steps, and the integrations between them.
A no-code automation platform connecting thousands of apps through simple trigger-action workflows. The most user-friendly entry point to marketing automation; less powerful than Make.com for complex logic.