Plain-English definitions of the technology, products and services used across retail and hospitality, from people counters and POS to cloud, cyber security and AI. 340 terms explained.
Browse by industry
Browse by topic
Jump to letter
Foot traffic, conversion and in-store analytics terms explained.
A people counter is a sensor that tallies people entering and leaving a store or venue, giving accurate visitor numbers for traffic and conversion analysis.
Footfall is the number of people who enter a store, centre or venue over a period, the base measure for judging traffic, marketing impact and conversion.
Conversion rate in retail is the percentage of store visitors who make a purchase, calculated by dividing transactions by footfall over the same period.
Dwell time is how long a person stays in a store, zone or queue, a useful signal of engagement, congestion and how well layout and service are working.
Occupancy monitoring tracks how many people are inside a space in real time, helping venues manage capacity limits, comfort, safety and staffing.
Queue management uses sensors and software to measure waiting lines and times, so stores can open registers or redeploy staff before customers give up.
A stereoscopic 3D sensor uses two lenses to measure depth and count people from above, giving accurate directional counts without capturing identifying images.
Thermal sensor counting detects people by body heat from overhead, giving reliable counts in any lighting without capturing a visible image of the person.
Video analytics uses software and AI to interpret camera footage automatically, turning video into counts, queue lengths, dwell times and movement data.
A heat map is a colour-coded overlay on a floor plan that shows where shoppers walk and linger, revealing the hot and cold areas of a store.
Traffic flow analysis studies the routes shoppers take through a store or centre, showing common paths and bottlenecks to improve layout and service.
Zone analytics measures visitor numbers and dwell time within defined areas of a store, showing which departments or displays attract and hold attention.
Average transaction value is total sales divided by the number of transactions, showing how much customers spend per purchase and how upselling is working.
Sales per square metre divides revenue by trading floor area, a retail productivity measure used to compare stores, test layouts and assess the value of space.
Staff scheduling optimisation matches rostered hours to expected customer demand, using footfall and sales data to reduce both queues and idle labour time.
Peak trading hours are the times of day and week when a store or venue is busiest. Learn how to find them and staff for them.
Capture rate is the share of passers-by who walk into a store. It shows how well a shopfront turns foot traffic into visitors.
Customer journey analytics follows how shoppers move through a store or across channels, showing where they stop, browse or leave.
Store benchmarking compares stores on measures like visitors, conversion and sales per square metre to find leaders and laggards.
People counter accuracy is how closely a sensor's visitor count matches the real number of people entering, and how to check it.
Bi-directional counting records both people entering and people leaving, so a counter can show live occupancy and true visitor totals.
Staff exclusion removes employees from visitor counts so footfall and conversion figures reflect customers only.
People counting privacy covers how counting sensors collect, store and use data without identifying shoppers, and what Australian rules may apply.
De-identified data has had personal details removed so individuals cannot reasonably be identified, which lets analytics proceed with lower privacy risk.
Sales per visitor is total sales divided by the number of people who entered, showing how much each visit is worth on average.
Labour cost percentage is total staff costs as a share of sales, a key measure of whether rosters match trading volume.
Sales per labour hour divides sales by hours worked, showing how productive a roster is across different times of day.
Wi-Fi tracking analytics estimates visitor numbers and movement from signals sent by customers' phones, with accuracy and privacy limits.
Like-for-like sales compare revenue from stores trading in both periods, removing the effect of openings and closures to show underlying growth.
Demographic analytics estimates broad characteristics of visitors, such as age band or gender, usually from cameras or loyalty data.
Queue abandonment is when customers leave a line or skip it entirely because the wait looks too long, costing the sale.
Footfall forecasting predicts how many visitors a store or venue will get, using past counts, seasonality, weather and events.
Staff-to-customer ratio compares how many customers are in a store with how many team members are on the floor.
A store analytics dashboard brings visitor counts, sales and staffing data into one screen so managers can spot trends and act on them.
Units per transaction is the average number of items in each sale, a measure of how well staff and layouts encourage add-on purchases.
A data retention policy sets how long a business keeps data, such as footage or customer records, and how it is deleted or de-identified.
Point-of-sale, payments, loyalty and hospitality ordering terms explained.
A point of sale (POS) system records sales, takes payment and updates stock at the moment a customer buys. Learn how POS works and what to consider.
Cloud POS is point of sale software hosted online and accessed through a browser or app, giving real-time sales and stock data across every store.
Multi-store management runs pricing, stock, staff and reporting across several locations from one central system. Learn how it works and what to plan.
ERPLY is a cloud point of sale and retail management platform covering POS, inventory and reporting. See what it does and how RetailCare supports it.
Nomni is a cloud retail platform combining POS, ecommerce and customer engagement tools. See what the category does and how RetailCare supports it.
Restroworks is a cloud restaurant platform covering POS, ordering and operations. See what it does and how RetailCare supports hospitality venues.
VenueSumo is a venue management and POS platform for pubs, clubs and large hospitality sites. Learn what it does and how RetailCare supports it locally.
EFTPOS (Electronic Funds Transfer at Point of Sale) lets customers pay by debit card at a terminal. Learn how it works in Australia.
Integrated payments link the payment terminal to the POS so the sale total passes across automatically, cutting errors and speeding reconciliation.
A payment terminal is the device that reads a customer card or phone and securely processes a payment. Learn the types and what to consider.
Contactless payments let customers pay by tapping a card, phone or watch on a terminal. Learn how NFC payments work and what retailers should consider.
PCI DSS is the global security standard for businesses that store, process or transmit card data. Learn what it requires and how to reduce exposure.
Payment surcharging was an extra fee on card payments. From 1 October 2026 the RBA ban covers eftpos, Mastercard and Visa cards.
Payment reconciliation matches sales recorded in your POS with card settlements and bank deposits to confirm every dollar is accounted for.
A loyalty program rewards repeat customers with points, discounts or perks and gives retailers buying data. Learn how it works and how to set one up.
A gift card is a prepaid stored value card or code redeemable at a business. Learn how physical and digital gift cards work and what to consider.
Omnichannel retail joins stores, website, app and marketplaces into one connected customer experience with shared stock and customer data.
Click and collect lets customers order and pay online, then pick up the goods from a store or venue. Learn how it works and common operational pitfalls.
A kitchen display system (KDS) shows orders on kitchen screens instead of printed dockets, helping staff prepare and track orders accurately.
QR code ordering lets diners scan a code at their table to view the menu, order and often pay from their phone. Learn how it works and what to consider.
A mobile POS lets staff take orders and payments on a tablet or handheld anywhere in the venue, removing the queue at the till.
A self-service kiosk lets customers browse, order and pay themselves, cutting queues and freeing staff for service.
A chargeback is when a cardholder disputes a payment with their bank and the funds are pulled back from the merchant.
Tap to Pay on phone turns a compatible smartphone into a contactless card reader, with no separate terminal required.
Buy now pay later lets customers split a purchase into instalments while the retailer is paid up front by the provider.
Online ordering lets customers place pickup or delivery orders through a website or app that feeds straight into the POS.
Tipping lets customers add a gratuity at payment. Modern POS and payment systems record and distribute tips accurately.
Table management software tracks bookings, seating and table status so a venue can turn tables quickly and serve more guests.
Least-cost routing sends a dual-network debit card payment down the cheaper network, usually eftpos, to reduce what a merchant pays in card fees.
The merchant service fee is what a business pays its payment provider to accept card payments, made up of interchange, scheme and provider margin.
A payment gateway securely passes card details from a website, app or POS to the bank network and returns an approval or decline.
Tokenisation replaces a card number with a random token so businesses can reuse or reference a payment without storing the real card data.
A pre-authorisation places a temporary hold on a customer's card for an expected amount, then captures the final amount once the sale is complete.
3D Secure adds an extra identity check, like a bank app approval or code, to online card payments to reduce fraud and shift chargeback liability.
Pay by link sends a customer a secure web link or QR to pay a set amount by card or wallet, without a terminal or being in store.
A GST tax invoice is the document a business gives a customer to support a GST credit claim; receipts must meet ATO rules on content.
Cash rounding adjusts cash totals to the nearest five cents because Australia has no one- or two-cent coins; card payments stay exact.
Cash up is the end-of-shift process of counting a till and matching cash and card totals to POS records to find variances.
A POS cash drawer is a lockable tray for notes and coins that opens automatically on cash sales via a signal from the receipt printer or POS.
A receipt printer is a thermal POS printer that produces customer receipts and kitchen dockets and often triggers the cash drawer.
A digital receipt is a sale record sent by email, SMS or app instead of printed, reducing paper and giving businesses a way to reach customers.
Offline mode lets a POS keep taking sales when the internet drops, storing transactions and syncing them once the connection returns.
An X report shows running sales totals without resetting; a Z report closes the trading period, records the totals and resets counters.
A void cancels an item or sale before payment completes; a refund returns money after a sale. POS controls and reporting help limit misuse.
Split bill lets a table or group divide one order into several payments by item, seat or share, each with its own tender.
Menu modifiers are options attached to a menu item, such as size, extras or cooking style, that change the price, prep notes or recipe.
Course firing lets servers send orders to the kitchen in stages, so entrees, mains and desserts are cooked and served at the right pace.
Order routing sends each item on an order to the right preparation point, such as the kitchen, bar or pizza station, by printer or screen.
Delivery platform integration feeds orders from apps like Uber Eats and DoorDash straight into the POS instead of a separate tablet.
A promotions engine applies discount rules, such as bundles, happy hours and vouchers, automatically at the POS and online.
Lay-by lets a customer reserve goods and pay in instalments, taking them once paid in full. Australian Consumer Law sets rules for it.
A house account lets approved customers charge purchases to an account and pay later by invoice, common for trade and corporate clients.
Recipe costing calculates the ingredient cost of each dish so a venue can price menu items to hit its target food cost and margin.
Menu engineering analyses each item's popularity and profit margin to decide what to promote, reprice, rework or remove from a menu.
A customer-facing display is a second POS screen showing the customer their items, total, offers and payment prompts as they are rung up.
POS user permissions control what each staff member can do, such as discounts, voids and refunds, with manager approval for sensitive actions.
Inventory, RFID, electronic shelf labels, signage and loss prevention terms explained.
Inventory management is how a retailer tracks and controls stock so the right products are available when needed without tying up too much cash.
A stocktake is a physical count of the stock in a business, compared against system records to correct discrepancies and measure shrinkage and accuracy.
A SKU is a unique code a retailer assigns to each distinct product variant, such as size or colour, so stock and sales can be tracked precisely.
A barcode is a machine-readable pattern of lines or squares that encodes a product identifier, so scanners can look up items quickly at the counter.
RFID uses radio waves to identify tagged items without line of sight, letting retailers count stock quickly and track individual items with high accuracy.
An RFID tag is a small label with a chip and antenna that stores a unique code and replies to an RFID reader, identifying the item it is attached to.
An RFID reader is a device that sends radio signals and collects the codes returned by nearby RFID tags, passing them to software to update stock records.
Electronic shelf labels are small digital price displays that update wirelessly from a central system, keeping shelf and checkout prices consistent.
An e-ink display is a low-power screen that looks like printed paper and uses energy only when its image changes, which suits electronic shelf labels.
ZKONG is a supplier of electronic shelf label systems that RetailCare supplies, installs and supports for retailers across Australia.
Digital signage uses screens to show managed content such as promotions, wayfinding or menus, and can be updated remotely across one or many locations.
A digital menu board is a screen that displays a venue's menu and prices, updated centrally so changes, specials and daypart menus go live instantly.
Loss prevention covers the policies, people and technology retailers use to reduce theft, fraud, error and other causes of lost stock and revenue.
Shrinkage is the difference between the stock a retailer should have and what it actually has, caused by theft, damage, error or fraud.
Electronic article surveillance uses tags on products and detection gates at exits to sound an alarm when unpaid items leave the store.
Video surveillance uses cameras and recording systems to monitor stores, deter theft and provide evidence when incidents occur.
Demand forecasting estimates how much of each product customers will buy over a future period, so retailers can plan stock, staffing and promotions.
AI inventory optimisation uses machine learning on sales and stock data to recommend what to order, where to hold it and when to move or mark down stock.
Overstock is holding more stock than demand justifies, while a stock-out is running out of an item customers want; both are costly inventory imbalances.
Replenishment is the process of topping up stock, from suppliers or other locations, so products stay available without ordering too much.
A cycle count checks a small section of stock regularly instead of counting everything at once, keeping stock records accurate.
A handheld scanner is a portable device that reads barcodes or RFID tags for receiving, counting, picking and price checks.
Dynamic pricing changes prices in response to demand, stock or time. Electronic shelf labels make it practical in store.
A purchase order is a formal order from a retailer to a supplier. Linking it to stock receipt keeps costs and inventory accurate.
A stock transfer moves inventory between stores or warehouses and records it in the system so counts stay correct.
Retail analytics turns sales, stock and customer data into reports and insights that guide buying, staffing and marketing.
Goods receiving is checking and booking in deliveries against a purchase order so stock counts and supplier invoices stay accurate.
Stock on hand is the quantity of an item the system says a store currently holds. It drives selling, ordering and reporting.
A par level is the target amount of an item to hold at the start of a period, so staff can order up to it and avoid running out.
A reorder point is the stock level at which a new order should be placed so stock arrives before the item runs out.
Safety stock is extra inventory held to cover unexpected demand or late deliveries so a product does not run out.
A planogram is a diagram showing which products go where on shelves and displays, so stores present ranges consistently.
Visual merchandising is how products are displayed in store and in the window to attract shoppers and encourage purchases.
A GTIN is the globally unique number behind a product barcode, issued through GS1 so any retailer can identify the same item.
A label printer prints barcode, price and product labels on adhesive stock, used for shelf edges, products and food prep.
A rugged mobile computer is a tough handheld device with a built-in scanner, used for stocktakes, receiving and picking.
Food waste tracking records what is thrown out and why, so kitchens can cut spoilage, over-production and ordering mistakes.
Food cost variance is the gap between what ingredients should have cost based on sales and what you actually used.
Temperature monitoring uses sensors and alerts to track fridges, freezers and hot-holding, protecting food safety and stock.
HACCP is a food safety system that identifies hazards and sets controls and checks, such as safe temperatures, to prevent food illness.
FIFO and FEFO are stock rotation methods: use the oldest stock first, or the stock that expires soonest first.
The cold chain is the unbroken, temperature-controlled path perishable goods follow from supplier to shelf or kitchen.
Lead time is the gap between placing an order with a supplier and having the stock received and ready to sell.
A warehouse management system tracks stock locations and directs receiving, picking and dispatch in a warehouse or stockroom.
Pick and pack is collecting ordered items from storage and packing them for dispatch to online customers or stores.
Batch and expiry tracking records the batch number and use-by date of stock, so you can rotate it and act fast on recalls.
A smart safe is a secure cash safe that counts and records deposits, cutting cash handling, counting errors and theft risk.
Exception-based reporting flags unusual transactions such as voids, discounts and refunds so managers can spot fraud and errors.
Markdown management is planning when and how far to cut prices on slow-moving or ageing stock to clear it with the best margin.
Inventory turnover measures how many times a business sells and replaces its stock in a period, showing how efficiently stock is used.
ABC analysis ranks products by their value to the business so the most important items get the closest stock control.
Managed IT, support, networking and cyber security terms explained.
A managed service provider (MSP) takes ongoing responsibility for an organisation's IT under a service agreement, covering monitoring, support, security.
A helpdesk is the single point of contact where staff report IT problems and requests, which are logged, prioritised and resolved by support technicians.
Remote support lets a technician diagnose and fix IT problems over a secure internet connection without travelling to site, often resolving faults quickly.
Onsite support is hands-on IT assistance where a technician attends a site in person to install, repair or replace equipment that cannot be fixed remotely.
A service level agreement (SLA) defines the measurable standards a provider commits to, such as response time, resolution time and availability.
Network infrastructure is the cabling, switches, routers, firewalls and Wi-Fi equipment that connect a business's devices to each other and the internet.
A Wi-Fi access point is a network device that connects wireless devices to a wired network, extending coverage across a store, venue or office.
SD-WAN is software that manages traffic across multiple internet links between sites, choosing the best path per application and simplifying networks.
VoIP (Voice over Internet Protocol) carries phone calls over the internet instead of traditional phone lines, enabling flexible cloud-based telephony.
A firewall is a security device or software that filters network traffic by rules, blocking unauthorised access while allowing legitimate communication.
Cyber security is the technologies, processes and habits that protect an organisation's systems, networks and data from attack, theft and disruption.
Endpoint protection is security software that defends computers, servers and mobile devices from malware, ransomware and suspicious activity.
Multi-factor authentication (MFA) needs two or more proofs of identity to sign in, such as a password plus an app code, so a stolen password is not enough.
Backup and disaster recovery combines copies of data with a tested plan to restore systems quickly after failure, theft, ransomware or disaster.
Managed printer services outsource the supply, monitoring, maintenance and consumables of a business's printers for predictable cost and less downtime.
IT consulting is independent advice on planning, selecting, implementing and improving technology so it supports business goals, budgets and risk needs.
Patch management is the process of testing and applying software updates to devices and systems so known security flaws and bugs are fixed promptly.
A VPN (virtual private network) creates an encrypted connection across the internet so users or sites can reach a private network securely.
The Essential Eight is the Australian Cyber Security Centre's set of eight baseline strategies to reduce cyber risk.
Ransomware is malware that locks or encrypts a business's data and demands payment. Backups and patching reduce the risk.
Phishing is a scam that uses fake emails or messages to trick people into giving up passwords, payments or access.
A business continuity plan sets out how a business keeps trading, or recovers fast, when systems, power or the internet fail.
Internet failover automatically switches a site to a backup connection, such as 4G or 5G, when the main internet drops.
RMM software lets an IT provider monitor, patch and fix devices remotely, often before staff notice a problem.
Network segmentation splits a network into separate zones, so a problem on one, like guest Wi-Fi, cannot reach payment systems.
The Notifiable Data Breaches scheme requires covered Australian organisations to notify people and the OAIC about breaches likely to cause serious harm.
The Privacy Act 1988 is Australia's main privacy law. It sets the Australian Privacy Principles for how covered businesses handle personal information.
NBN business internet is internet delivered over the National Broadband Network with business-focused plans, support and sometimes service guarantees.
4G/5G fixed wireless internet delivers broadband over a mobile network through a modem, used as a primary or backup connection for stores.
Structured cabling is the organised, standards-based system of data cables, outlets and patch panels that connects a venue's devices to its network.
Guest Wi-Fi is a separate wireless network for customers, kept apart from POS and staff systems and often protected by a splash page.
A Wi-Fi site survey measures signal strength, interference and capacity in a venue to plan where access points should go.
Power over Ethernet sends electrical power and data over one network cable, so devices like cameras and access points need no separate power outlet.
A network switch connects wired devices such as tills, printers and access points inside a venue and directs data between them.
A business router joins a store's network to the internet and manages traffic, security and failover, unlike a basic home modem.
Mobile device management lets a business remotely configure, secure and update tablets, phones and handhelds, including locking or wiping lost devices.
Single sign-on lets staff use one secure login for many business applications, making access easier to manage and to switch off.
A password manager stores and fills strong, unique passwords for every account, so staff do not reuse passwords or write them down.
Least privilege means giving each person and system only the access needed for their job, limiting damage from mistakes and compromised accounts.
The 3-2-1 rule means keeping three copies of data on two types of storage with one copy offsite, so a single failure or attack cannot destroy everything.
RTO is how long a system can be down and RPO is how much data loss is acceptable. Together they set backup and recovery requirements.
Hardware lifecycle management plans when tills, PCs, network gear and printers are bought, maintained, replaced and safely retired.
A UPS is a battery unit that keeps tills, routers and servers running through short power cuts and protects them from surges.
A thermal receipt printer prints receipts, dockets and kitchen tickets by heating special paper, with no ink or toner required.
A cloud phone system delivers business calling over the internet, with features like call routing, voicemail and mobile apps, and no on-site phone exchange.
Business email compromise is a scam where criminals impersonate a boss or supplier by email to trick staff into paying money or sharing data.
Security awareness training teaches staff to spot scams, handle data safely and report problems, reducing the chance of a successful attack.
Penetration testing is an authorised simulated attack on systems or networks to find weaknesses before real attackers do.
An incident response plan sets out who does what, and in what order, when a cyber attack or data breach is discovered.
Mean time to resolve is the average time it takes an IT provider to fully fix an issue from when it was reported.
Ecommerce, integration, cloud, SEO and digital marketing terms explained.
Ecommerce is the buying and selling of goods or services over the internet. Learn how online stores work and how they connect to POS and stock.
Headless commerce separates the customer-facing website from the ecommerce back end, linked by API, for flexible, fast and multi-channel experiences.
An API is a defined way for one software system to request data or actions from another. See how APIs connect POS, ecommerce and other retail systems.
System integration links separate business systems so they share data automatically. Learn how POS, ecommerce, stock and finance systems are connected.
A webhook is an automatic message one system sends to another when an event happens, such as a new order, removing the need to keep checking for updates.
Cloud computing delivers servers, storage, databases and software over the internet on demand, so businesses avoid owning and maintaining hardware.
Amazon Web Services is Amazon's cloud platform, offering hosting, databases, storage, security and AI services that retailers use to run applications.
SaaS, or software as a service, is software you access online by subscription instead of installing and maintaining it yourself on your own servers.
A mobile app is software built to run on smartphones or tablets. Retailers use apps for loyalty, ordering, staff tools and a direct channel to customers.
A progressive web app is a website built with modern web technology so it can be installed, work offline and feel like a native mobile app.
Digital marketing promotes a business through online channels such as search, social media, email and advertising, and measures results through data.
SEO, or search engine optimisation, improves a website so it appears higher in unpaid search results and attracts more relevant visitors.
Local SEO helps a business appear in map and search results for nearby customers, using listings, reviews and consistent location details.
GEO is the practice of shaping content so AI assistants and AI search tools can understand, trust and cite it when generating answers.
Structured data is standardised code on a web page that tells search engines and AI tools what the content describes, such as products or opening hours.
Virtual reality training uses headsets and simulated scenarios so staff can practise tasks safely and consistently without disrupting trading.
A digital wallet stores payment cards, passes, loyalty cards and tickets on a phone, letting customers pay or redeem offers without physical cards.
Google Analytics 4 is Google's event-based web and app analytics tool, showing how visitors find and use a site and which actions lead to sales.
A customer data platform collects customer data from many systems into one profile per person, so marketing and service use a consistent single view.
Pay-per-click is online advertising where you pay each time someone clicks your ad, such as in Google search results.
A Google Business Profile is the free listing that shows your business details, reviews and photos in Google Search and Maps.
PIM is a central system for product descriptions, images and attributes, feeding the website, POS and marketplaces consistently.
CRO is improving a website so more visitors complete a goal, such as buying or sending an enquiry.
Marketplace integration connects your stock and orders to online marketplaces so sales from each channel stay in sync.
Abandoned cart recovery follows up shoppers who add items to an online cart but leave without paying, usually by email or SMS.
Email marketing sends promotional and service emails to customers who have opted in, such as offers, newsletters and receipts.
SMS marketing sends short promotional or service text messages to customers who have agreed to receive them.
Marketing automation uses software to send timed, personalised messages triggered by customer actions, without manual sending.
Online reputation management is monitoring and responding to reviews and mentions so a business keeps a trusted public image.
Social commerce lets customers discover and buy products inside social media apps such as Instagram, Facebook and TikTok.
Web accessibility means designing websites so people with disability can use them, guided by the WCAG standard.
Core Web Vitals are Google's measures of how fast, responsive and visually stable a web page feels to real visitors.
A product feed is a structured file of product data sent to Google, social and marketplaces so items can be listed and advertised.
Personalisation tailors offers, messages and product suggestions to each customer based on their behaviour and preferences.
First-party data is information a business collects directly from its own customers, such as purchases, sign-ups and preferences.
Cookie consent is the notice and choice a website gives visitors about cookies and tracking tools used for analytics and advertising.
Conversion tracking records when a visitor completes a valuable action, such as a purchase or booking, and where they came from.
UTM parameters are tags added to a link's URL so analytics can show which campaign, channel or email sent each visitor.
A/B testing shows two versions of a page, email or offer to different visitors to see which performs better.
Third-party delivery platforms are marketplaces such as Uber Eats and DoorDash that list a venue's menu and deliver orders for a commission.
Google Tag Manager is a free tool for adding and managing analytics and advertising tags on a website without editing its code each time.
The Spam Act 2003 requires Australian businesses to get consent, identify themselves and offer an unsubscribe when sending commercial messages.
A content delivery network stores copies of a website's files on servers around the world so pages load faster for visitors.
Artificial intelligence, AI agents and automation terms explained for retail, QSR and hospitality.
An AI agent is software that uses a language model to pursue a goal, deciding steps and using tools such as business systems to complete tasks.
A chatbot is software that converses with people by text or voice to answer questions or complete simple tasks, from rule-based scripts to AI models.
A large language model is an AI trained on vast amounts of text to understand and generate language, powering modern chatbots.
Machine learning is software that learns patterns from data to make predictions, such as forecasting demand or spotting fraud.
Conversational AI lets customers talk or type to a virtual assistant that answers questions and completes simple tasks.
Agentic AI describes AI systems that pursue a goal by planning steps, using tools and acting, rather than only answering a single question.
Tool use lets an AI model call defined software functions, such as checking stock or booking a table, instead of only generating text.
Model Context Protocol (MCP) is an open standard that lets AI assistants connect to business tools and data in a consistent way.
Workflow automation uses software to run repeat business tasks automatically from a trigger, cutting manual data entry and follow-up.
Robotic process automation (RPA) uses software bots to copy a person's clicks and keystrokes in existing applications to do repetitive tasks.
Human-in-the-loop means a person reviews, approves or corrects AI output at key points before it affects customers or money.
AI guardrails are the rules and technical controls that keep an AI system within safe, on-brand and approved behaviour.
AI orchestration is the coordination layer that decides which model, tool or step runs next so a multi-step AI task finishes reliably.
A multi-agent system uses several specialised AI agents that share a task, such as one for bookings and another for returns.
An AI copilot is an assistant built into the software staff already use, suggesting actions and drafting content while the person stays in control.
An AI voice agent answers and makes phone calls in natural speech, handling bookings, orders and questions without a human on the line.
An AI receptionist answers calls and messages for a business, giving information, taking bookings and passing on messages around the clock.
AI customer service uses chat or voice AI to answer customer questions, track orders and resolve routine issues, escalating complex cases.
A lead qualification bot chats with website visitors to collect their needs and contact details, then passes suitable leads to sales.
Human handoff is the point where an AI assistant passes a conversation to a staff member, ideally with the full context attached.
Intent recognition is how an AI works out what a customer is trying to do, such as book, return or ask about stock, from their words.
An AI knowledge base is the approved content, such as menus, policies and FAQs, that an AI assistant draws on to answer accurately.
Conversation design is the craft of planning how an AI assistant talks, from tone and wording to how it recovers from confusion.
An AI chat widget is the chat window on a website where visitors talk to an AI assistant that answers questions and captures enquiries.
AI order taking uses voice or chat AI to capture customer orders, confirm them and send them to the POS or kitchen.
An AI scheduling assistant books, moves and cancels appointments or tables by conversation, checking live availability.
AI email drafting uses AI to write or suggest email replies and messages, which staff review and send.
Assistive AI suggests while a person decides; autonomous AI acts on its own within set limits. Most businesses use a mix.
No-code automation lets non-programmers connect apps and build automated workflows using visual tools instead of writing code.
Connecting AI to your POS and CRM lets it use live sales, stock and customer data to answer questions and take actions.
A token is the small chunk of text, roughly a short word or word fragment, that an AI model reads and writes, and the unit it is priced by.
A prompt is the text instruction, question or context you give an AI model so it knows what to do and how to respond.
Prompt engineering is the practice of wording and structuring AI instructions so a model gives accurate, consistent, usable answers.
A system prompt is the hidden set of standing instructions that tells an AI assistant its role, tone, rules and limits before any customer speaks.
The context window is the maximum amount of text, measured in tokens, an AI model can consider at one time, including your prompt and its reply.
An embedding is a list of numbers that represents the meaning of text or an image, so AI can find items that are similar in meaning.
A vector database stores embeddings and quickly finds the entries closest in meaning to a question, powering AI search and retrieval.
RAG lets an AI look up relevant information from your own documents first, then write its answer from what it found, reducing made-up replies.
Fine-tuning is further training of an existing AI model on your own examples so it adopts a specific style, format or specialised behaviour.
An AI hallucination is a confident-sounding answer that is wrong or invented, such as a made-up price, policy or product feature.
Multimodal AI can understand and produce more than one type of content, such as text, images, audio and video, in the same model.
Inference is the step where a trained AI model is actually used to produce an answer or prediction from new input.
Temperature is a model setting controlling how predictable or varied an AI's answers are, from consistent and factual to creative.
A small language model is a compact AI model that is cheaper and faster than the biggest models and good enough for focused business tasks.
A foundation model is a large, general-purpose AI model trained on broad data that can be adapted to many different tasks.
A neural network is a computing system of connected layers, loosely inspired by the brain, that learns patterns from examples.
Deep learning is a type of machine learning that uses many-layered neural networks to learn complex patterns from large amounts of data.
Natural language processing is the branch of AI that lets computers read, interpret and produce everyday human language.
Training data is the collection of examples an AI model learns from, and its quality and coverage shape how well the model performs.
Grounding means tying an AI's answers to trusted, specific sources such as your menu or policies so it does not rely on guesswork.
An AI model provider is a company that builds and hosts AI models, which businesses use directly or through software products.
An open-source or open-weight AI model has its files released publicly so businesses can run and adapt it on their own infrastructure.
AI latency is the delay between a person asking an AI something and getting the answer, which matters for live customer and voice use.
AI usage cost is what it costs to run AI, usually priced by tokens processed, so model choice and message length drive the bill.
AI evaluation is testing how well a model or assistant performs, using standard benchmarks and your own real-world test questions.
Responsible AI means designing and using AI fairly, safely and accountably, with people kept in charge of the outcomes.
AI bias is when an AI system gives systematically unfair or skewed results because of the data it learned from or how it was built.
Explainable AI means being able to understand and describe, in plain terms, why an AI system produced a particular result.
AI and privacy covers how personal information is collected, used and protected when AI tools process customer or staff data.
The Australian Privacy Principles are 13 rules under the Privacy Act 1988 on how covered organisations handle personal information.
The Australian Government's Guidance for AI Adoption sets out six practices for using AI safely and responsibly in business.
Data governance is the set of rules and responsibilities for how a business collects, stores, uses and protects its data.
Data quality is how accurate, complete, consistent and up to date your data is, and it largely determines how well AI performs.
Synthetic data is artificially generated data that mimics real data, used to test or train systems without exposing real customers.
Model drift is when an AI system gets less accurate over time because the real world has changed since it was trained.
AI literacy is the basic knowledge staff need to use AI tools well, spot their limits and handle data and risks sensibly.
A deepfake is AI-generated fake audio, image or video that convincingly imitates a real person, often used in scams.
Prompt injection is an attack that tricks an AI tool into ignoring its instructions by hiding commands in text it reads.
Shadow AI is staff using AI tools that the business has not approved or reviewed, often with company or customer data.
An AI acceptable use policy is a short document setting out which AI tools staff may use, for what, and with what data.
AI vendor due diligence is checking an AI supplier's security, data handling, reliability and terms before you buy.
Data residency is where in the world your data is physically stored and processed, which affects the laws that apply to it.
AI transparency means being open with customers and staff when AI is used, so people know when they are dealing with a machine.
Copyright and AI-generated content covers who owns AI-made text and images and the risk of copying others' work.
An AI audit trail is a record of what an AI system was asked, what it did and who approved it, kept for review and accountability.
AI ROI compares the benefits of an AI project, such as time saved or sales gained, against its full costs.
An AI pilot is a small, time-limited trial of an AI tool in a real business setting to test value and risk before wider rollout.
Change management for AI is helping staff understand, trust and use new AI tools so the rollout actually sticks.
Australian Consumer Law applies to AI-driven advertising, chat and pricing in the same way as to anything else a business says or does.
A recommendation engine uses past behaviour and product data to suggest items a shopper is likely to want, lifting basket size and repeat sales.
Retail personalisation tailors offers, content and product displays to each shopper using their data, so the experience feels relevant rather than generic.
Computer vision is AI that interprets images and video, letting cameras count people, read shelves, spot items and detect events automatically.
Visual search lets shoppers find products by uploading or snapping a photo instead of typing, using AI to match look, shape and colour.
Markdown optimisation uses AI to decide when and how much to discount slow-moving stock so it sells through with the least lost margin.
AI voice ordering uses speech recognition and language models to take customer orders by voice at drive-thrus, phones and kiosks.
Virtual try-on lets shoppers see how clothing, eyewear, makeup or furniture would look on them or in their space using a phone, mirror or website.
Sentiment analysis uses AI to read reviews, surveys and messages and classify them as positive, negative or neutral, revealing trends at scale.
Predictive rostering uses forecast demand to suggest how many staff you need each hour, matching labour to sales while respecting award rules.
Payment fraud detection uses AI to score transactions for risk in real time, blocking or flagging suspicious orders and card payments.
AI shoplifting detection uses cameras and software to flag suspicious behaviour or checkout errors so staff can act before stock leaves.
AI product tagging reads product images and text to fill in attributes like colour, material and style, making catalogues searchable and consistent.
AI product descriptions are written by generative AI from specs and images, helping retailers publish consistent listings quickly.
Demand sensing uses fresh signals such as weather, events and current sales to adjust short-term forecasts, rather than relying on history alone.
AI menu engineering analyses sales and costs to show which dishes earn the most and how to price, position and promote them.
Food waste forecasting predicts how much to prepare and order each day so kitchens cut waste and avoid running out of popular items.
Customer segmentation groups shoppers by behaviour, value or needs so each group gets more relevant offers and service.
Churn prediction uses AI to spot customers who are likely to stop buying, so you can reach out before they are gone.
Next-best-offer picks the single most relevant offer or message for each customer at each moment, based on their data and business goals.
AI marketing content uses generative AI to draft emails, social posts, ads and images, speeding up campaigns for small teams.
AI review responses draft replies to Google and social reviews so businesses reply quickly and consistently, with a person approving.
AI site search understands what shoppers mean, not just the words they type, returning better results from your online catalogue.
A smart shelf uses sensors, cameras or RFID with AI to track stock on the shelf, spot gaps and flag misplaced or mispriced items.
Autonomous checkout lets shoppers pick up items and leave, with cameras, sensors and AI tracking what they take and charging them automatically.
AI energy management optimises lighting, cooling and refrigeration to cut power bills without hurting comfort or food safety.
Tell us what you would like explained and our team will add it, or talk to us about your own retail technology.
Contact RetailCareTell us what you need and the RetailCare team will get back to you. Your enquiry goes straight to our sales team.