Biky dictionary
The words you will run into all over the platform and throughout this documentation, explained without assuming anything. Each one links to the page where it is covered in depth and configured.
The dashboard acronyms — TPR, TTE, MQL, SQL, CSAT — are under Metrics.
If something doesn't click, start with Space, Template, SalesCard and KB System: they are the four terms that come up the most, and almost everything else depends on them.
Conversations
AI agent
The automatic salesperson that handles your conversations. In the platform the module is called Biky admin, but the agent is not always named Biky: you give it the name, the face and the way of speaking you want. It handles things on its own until a human is needed, and then it transfers.
Smart Chat
The unified inbox: every conversation from every channel on one screen, whether the person wrote on WhatsApp, Instagram or your site's chat.
Space
A sectioned inbox: an inbox of its own for an area or a location — Sales, Customer service, the Medellín branch. Every conversation lands in a space, and each person on the team sees only the spaces assigned to them. It is the mechanism that splits up the work: when someone says nothing shows up for them, it is almost always that they have no spaces assigned.
Each space has a color, and that color tints the channel icon in the chat list, so in the all spaces view you can tell at a glance which one a conversation belongs to.
→ Spaces
Space group
One level above the space, for when you have many: Sales Bogotá, Sales Medellín and Sales Cali grouped under Sales. With two or three spaces you don't need it.
Template
A WhatsApp message approved by Meta. WhatsApp does not let you write to someone first with free text: you have to use a message Meta reviewed beforehand. That is why you need templates to start a conversation and for any campaign.
Approval is not instant: you create it, submit it for review, and it sits as pending until Meta answers.
24-hour window
The WhatsApp rule behind templates: from the moment a contact writes to you, you have 24 hours to answer with whatever you want. After that, you can only pick the conversation back up with an approved template.
Proactivity
Writing to someone first instead of waiting for them to write. It is what Outbound campaigns, flows and the start-a-chat button do — and why they all depend on a template.
Transfer
Handing a conversation to another person or another space, with the whole history. It is what the agent does when a case is beyond it, and what an advisor does when the customer belongs to another area.
Typification
The reason a conversation was closed: Sale made, Not interested, General information. It is picked when closing the chat, and it is what gets counted later in Metrics to know where the team's time is going.
Quick reply
A canned message the team inserts by typing / in the text box. It is for what repeats every day: prices, hours, the payment link.
Contacts
CDP
Customer Data Platform: the contact database. Everything you know about each person — their details, their conversations, their purchases — lives there, and that is where the lists you send campaigns to come from.
→ CDP
Contact
A person (or a company, or a branch) with a record in the CDP. It is created on its own the moment someone writes through any channel, and it can also be created by hand or imported from a file.
Entity
The kind of record: People, Companies or Branches. Each entity has its own fields and its own segments; they are separate databases.
→ Contacts
Lead
A contact who showed interest and has not bought yet. It is not a formal platform state: it is the word used for the contact who comes in through a form, an ad or the widget, and who starts a SalesCard.
Segment
A named subset of the database, so you don't have to rebuild the same filter every time. It is what you pick as the audience of a campaign.
Dynamic segment
It saves the question, not the list. "Contacts in Bogotá who accepted data consent": every time you open it, the CDP asks that question again and shows you who meets it today. New contacts who meet it come in on their own, and those who stop meeting it drop out on their own.
→ The difference between dynamic and static
Static segment
It saves the exact list of contacts, as it stood the day you imported it. It does not grow or shrink. It is what you want for something closed — attendees of the event on the 15th — and what you don't want for a recurring campaign.
Contact fields
The data each record holds. Besides the usual ones (name, phone, email), your account has custom fields specific to your industry, shown in a section named after your KB System.
KB System
The industry category Keybe assigns to your account when it creates it: automotive, health, fashion, real estate, education, retail… It is not cosmetic. It determines:
- the custom fields on the contact card and the name of that section (Car, Health…);
- the default funnels and steps that come ready-made and cannot be modified;
- the list of positions you can assign to your team;
- in the automotive sector, the distinction between a digital and a floor advisor.
Keybe configures it; it is not changed from the platform. If something about your operation doesn't fit, the adjustment is requested from your Keybe contact.
Sales
Funnel
The sales pipeline: the series of stages an opportunity goes through, from the moment the interested party arrives until they buy or are lost. You can have several, one per line of business.
→ Funnels
Step
Each stage of a funnel — Lead, Quote sent, Closed won. Moving a card from one step to the next is what records that the sale advanced.
→ Configuring funnels and steps
SalesCard
The card for a sales opportunity: a contact, a product, an amount, an owner and the funnel step it sits in. It is the unit the sale is measured in — a conversation may lead to none, and one contact can have several.
Owner
The person in charge of a SalesCard. It is not the same as whoever handles the conversation: anyone in the space can take the chat, but the sale has an owner.
Temperature
How hot the opportunity is — Cold, Warm, Hot, Urgent. A person sets it by hand.
SLA
How long the SalesCard has gone without anyone touching it. It is calculated on its own, and when it reaches Critical that is the sign the opportunity is going cold through neglect.
Channels
Channel
Where a message comes in or goes out: WhatsApp, Instagram, Facebook Messenger, TikTok, your site's chat, a form. There are ten in total, and each one connects differently.
Line
A WhatsApp number connected to the platform. An account can have several, and each one is assigned to a space. It has its own display name, its own messaging limit and its own quality.
Line alias
The short name your team — and sometimes the visitor — identifies a line by. When a form or a widget offers several lines, what the person sees is the alias, so it had better say something: Sales Bogotá, not Line 2.
Messaging limit
How many new conversations you can start per day: 1K, 10K, 100K. Meta assigns it and it only goes up if you send good messages. The important part, and the one that surprises everyone: it is shared across all the lines of the same business, so splitting a campaign across two lines does not double your reach.
→ The limit: what that 10K means
Line quality
The grade Meta gives your number based on how the people receiving your messages react. If a lot of them block or report you, it drops; if it drops too far, Meta cuts your limit or suspends the line.
Display name
The name the person sees in their WhatsApp when you write to them. Meta reviews it before approving it and it cannot be just anything: it has to match your business.
API key
The identifier the code for a widget, a form or the web chat is generated with. It is what connects that piece of your site to your account.
Meta forms
The lead forms on Facebook and Instagram ads. Once connected, every person who fills one in comes into the platform as a contact and can trigger a template or create a SalesCard, without anyone downloading a leads file.
Automation
Flow
An automation: "when this happens, do that". It is built by dragging blocks on a board, without programming.
→ Flows
Trigger
The "when" of a flow: the event that sets it off — a new message arrives, a SalesCard moves, an appointment reminder fires. A flow has exactly one.
Node
Each block of the flow: send a message, wait, ask, branch on a condition. They are chained with arrows.
→ Nodes
Intention
What the agent detects the person wants — wants a quote, asking about warranty, wants a human — so it can answer differently in each case. You define them.
Skill
Something the agent can do beyond talking: look up the catalog, book an appointment, transfer to an advisor, send a voice note. Each one is switched on or off, and some have settings of their own.
→ Skills
Knowledge
What the agent knows about your business: the documents, links and text you load so it answers with your information and not with generalities.
Simulator
A test conversation with your agent, without spending a real conversation or writing to a customer. It is where a change gets checked before it goes live.
Campaigns
Outbound
The outgoing messages module: campaigns over WhatsApp, email and SMS to a segment of your database.
→ Outbound
Campaign
One send, to one list, at one moment, with one message. It is put together by choosing the channel, the segment and the template.
Bulk send
What a campaign looks like in the inbox: many chats opened at once that nobody has to handle yet. That is why it is worth separating them from the All filter, and having them return to the inbox only when the person replies.
→ Spaces
Metrics
The dashboards' vocabulary, which is the densest in the platform and the most misread: nearly every argument about a number comes from two people not measuring the same thing.
TPR
Average response time: how long the advisor takes to reply, averaged over every reply in the period. It is not the time to the first reply: it is all of them.
TTE
Total wait time: the sum of the time that advisor's customers spent waiting. Unlike TPR, which is an average, this one is a total: it grows with volume. A good TPR with a huge TTE means they answer fast, but for a lot of people.
S/R
No reply. It appears where a time would go when there was no reply to measure. It is not a zero, it is an absence: that advisor answered nothing in the period.
MQL
Marketing Qualified Lead: the contact who came in and showed initial interest. It is the first step of the funnel.
SQL
Sales Qualified Lead: the contact who also qualified as a real sales opportunity. It is the denominator of the conversion rate, which is why the MQL → SQL jump is the one watched most closely.
CSAT
Customer Satisfaction: how satisfied the customer is, measured with surveys when the conversation closes. It only fills in if you have surveys running; if the CSAT blocks come up empty, that is why and not a bug.
SKU
The code that identifies each product in your Catalog. In Products of interest it is what tells two variants of the same product apart.
Global score
A mark from 0 to 100 summing up the advisor's performance. The rating is the label that number earns:
| Score | Rating |
|---|---|
| 100 | Excellent |
| 90 – 99 | Very good |
| 80 – 89 | Good |
| 70 – 79 | Acceptable |
| 60 – 69 | Regular |
| 45 – 59 | Poor |
| 30 – 44 | Bad |
| under 30 | Critical |
The cuts are deliberately uneven: the difference between 85 and 95 weighs more than between 35 and 45, because below 60 there is a problem to deal with anyway.
Top performer
The advisor with the highest score in the period, with their mark. It is a figure for the filtered period, not a historical one: changing the date range can change who it is.
Interactions
The advisor's exchanges, split three ways: total, answered and unanswered. The column that matters is the third.
% success
The share of interactions that did get answered. Watch out for small numbers: 0% out of one interaction and 50% out of four are not comparable, whatever the percentage suggests.
Agent empathy
The block that brings together TPR, TTE, interactions and rating per agent. The name comes from Keybe's Empathy Protocol, linked from the block itself: what it measures is how well attended the customer felt, not how much the advisor sold.
Session
One conversation handled, start to finish. The same person can generate several sessions in the period if they write on different days.
Online time and working time
Online is the minutes the advisor had the platform open and available. Working is, of those, the ones actually spent handling conversations — sending messages. The table shows it as a share of online time, and that is where it gets interesting: eight hours online with one hour working says something very different from two hours online with an hour and a half working.
Average duration
How long a session lasts on average. It goes up both when conversations are handled thoroughly and when they are left open without closing, so it is read alongside open sessions.
Elapsed time
How long a lead has gone without moving from the step it sits in (in some blocks, hours in step). It is the figure that exposes dormant opportunities: 24d 3h in the same step is not an opportunity, it is an overdue task.
Origin
Who spoke first in that session. There are four:
| Origin | Who started it |
|---|---|
| Inbound | The customer. They wrote in. |
| Outbound | An Outbound campaign. |
| Automation-initiated | A flow, set off by an event. |
| Agent-initiated | Someone on the team, who started the chat by hand. |
The mix says a lot: a 100% inbound operation is not going looking for customers, and a 100% outbound one is not being found.
Reasons for not advancing MQL to SQL
Why a lead stalled at the door. The platform classifies them like this:
| Reason | What happened |
|---|---|
| No conversation | The lead never had a chat linked to it. It came in as a record and nobody wrote to them. |
| Customer didn't reply | There is a chat, but the customer never answered a single message. |
| Send failed | The customer didn't reply, and on top of that the WhatsApp message never arrived. |
| Replied but didn't advance | There was a conversation, and the lead still didn't move a step. |
| Possible duplicate | The same contact already has a more recent card in the funnel; most likely the team is working that other one. |
The distinction earns its keep because each one calls for a different action: no conversation is a problem with your team, customer didn't reply with the message, send failed is technical, and replied but didn't advance is with the offer.
Source
Where the lead came from: organic, web, agency, a campaign.
Campaign, ad set and ad
Meta's three levels, from broadest to most specific. They let you know not just that a campaign brought leads, but which ad brought them.
Cost per lead
What was spent divided by the leads it brought. It is the number you compare one paid channel against another with.
Mention
Every time a customer names something in a conversation — a product, a model, a term — whether or not they buy. It is what people ask about, which is not always what they buy.
The mentions vs. unique users block tells how many times something was named apart from how many people named it: a hundred mentions from five people is a conversation; a hundred mentions from eighty people is a trend.
Products of interest
The products asked about most, with their SKU and their number of questions. A product with many questions and few sales is badly explained or badly priced.
Previous period
What each indicator is compared against: the immediately preceding period, the same length as the one you picked. Eleven days compare against the previous eleven, not against last month. To compare whole months you have to pick whole months.
Scoreboard
Scoreboard
The month's board: whether you are going to hit the goal and what to do if you aren't. Unlike Metrics, which looks back, everything here reads against what is left.
Signal of the day
The board's one sentence: what to do today, drawn from crossing your goal with your progress and with the funnel step losing the most. It is generated on its own, every day.
Snapshot of the day
The board's numbers are not recalculated on every click: they are read once and stay put, with the time on display. That is what lets you look at them as a group without them shifting while you talk.
Industry standard
The percentage each funnel step should convert at in your industry — 80% of MQLs become interested, 50% of appointments reach show. It is what the board measures a broken step against, and the basis for calculating the goal backwards.
Goal
How much you want to sell this month. You write one figure, the invoices one; the rest of the chain — pre-sales, shows, appointments, SQL, MQL — is calculated backwards with the industry standard.
Quota
The share of the goal each person on the team carries. It is split evenly by default; pinning one by hand redistributes the rest among the unpinned so the sum still comes to exactly the goal.
BIKY Token
The unit measuring the text the board's AI processes: about 4 characters, three quarters of a word. It counts what it reads and what it writes, and is independent of your WhatsApp conversations' consumption.
OEM
The brand's manufacturer. In the consolidated view it is one of the filters, alongside country, group and brand, for looking only at the dealerships of one automaker.
Billing
Plan
What your account has contracted per month: a number of AI chats included and a monthly cost. The consumption bar in Billing shows how many of that total you have used.
Tier
Your account's platform level. It has its own name, label and cost, and depends on your KB System: a dealership's tiers are not a clinic's, because they don't include the same things.
It shows up in the product names on your invoice — Plan 500 Tier 1, Plan 3,000 Tier 4 — and it is the part of the price that does not depend on how many conversations you have.
Bundle
That the plan is priced as a single package: the platform tier plus the conversation volume, at one price. That is why changing plans shows you one total and not two lines: the only part known separately is the tier cost, and what is left is the Biky AI share.
Overflow
Consumption beyond what your plan includes. It does not stop the operation as long as you have balance: it is billed separately, at its own price, which also depends on the tier (Overflow Tier 1, Overflow Tier 4) or on the product (Overflow Biky Chat).
AI chats and AI messages
The two units the agent consumes, contracted separately. A chat is one conversation; a message is each exchange inside it. When you change plans you pick the volume of each.
Package
A block of consumption reserved for the next billing cycle: SMS, AI chats, AI messages. Change package does not touch the current cycle — it picks the one that activates in the next.
Subscription
A recurring charge on the account. It can be cancelled, and then it stops renewing from the next cycle on; what is already paid stays valid until the cutoff.
Billing day
The date your cycle closes and the invoice is issued. It is the reference for everything else: this cycle, last cutoff and the moment reserved packages activate.
Balance
The money available in the account. Without balance you cannot take conversations, send campaigns or send templates — so when the team reports they "can't take chats", this is the first thing to look at.
→ Billing
Account and team
Type
A person's role in the account — Agent, Coordinator, Trainer, Admin, Account Manager, Owner — and what decides what they see and what they can do. It is the first thing to check when someone "can't see anything" or sees too much.
→ Type: what each person can do
Position
The person's actual position in your company — Sales Advisor, Brand Manager. It grants no permissions: it is descriptive, and the list of options comes from your KB System.
→ Position
Digital / floor advisor
In automotive accounts, whether the person works over chat (digital) or in person at the dealership (floor). It is there so you compare like with like in Metrics: a floor advisor's response time is not measured the same way as a digital one's.
Branch
A physical location of the business. The agent uses it to answer about addresses and hours, and it can be filtered by space.
→ Branches