Reference
59 terms defined the way an operator would explain them on a call — what it means, when it matters, and where teams get it wrong. Written to be quotable in isolation, because that is how people and AI assistants read definitions.
Terms defined
59
Categories
7
Implementations
50+
Where these definitions come from
Client retention
92%
01
10 terms
A contact whose fit and behaviour meet the agreed threshold for sales follow-up.
An MQL is a contact that marketing has judged ready for sales contact, based on an agreed combination of fit (job title, company size, industry) and behaviour (form fills, pricing page visits, demo requests). The definition only works when sales agrees to it in writing and the threshold is derived from historical conversion data rather than opinion. In HubSpot, MQL is a lifecycle stage value, and the date a contact entered it is stamped on the record — that stamp, not the contact create date, is what reporting should filter on.
A lead that sales has worked and confirmed as a genuine opportunity to pursue.
A Sales Qualified Lead is a contact that a sales rep has engaged with and qualified as worth pursuing, usually against criteria such as budget, authority, need and timing. It sits after MQL and SAL in the lifecycle. The practical distinction is who did the qualifying: marketing qualifies an MQL from data, sales qualifies an SQL from a conversation. Tracking the MQL-to-SQL conversion rate is the single fastest way to see whether a scoring model reflects reality.
An MQL that sales has formally accepted into their queue for follow-up.
A Sales Accepted Lead is the handshake stage between MQL and SQL: sales has looked at the lead, agreed it meets the criteria and accepted responsibility for working it. Introducing SAL is the cleanest way to end arguments about lead quality, because rejection becomes an explicit, measurable event with a reason code rather than silence. A healthy funnel accepts most MQLs; a low acceptance rate is a scoring problem, not a sales attitude problem.
The standard HubSpot property tracking how far a contact or company has progressed.
Lifecycle stage is HubSpot's built-in property describing where a record sits in the buying journey: subscriber, lead, MQL, SAL, SQL, opportunity, customer, evangelist or other. It should only ever move forward automatically; backward movement should require deliberate action. Because HubSpot stamps the date a record enters each stage, lifecycle stage is the backbone of accurate funnel reporting — provided every stage has a written definition everyone applies the same way.
A step in a sales pipeline, defined by a buyer action rather than a seller feeling.
A deal stage represents a step in the sales process. Good stages are defined by verifiable buyer actions — 'demo completed', 'proposal sent', 'security review passed' — rather than internal sentiment like 'interested'. Each stage carries a win probability used in forecasting, so stages defined by feeling produce forecasts built on feeling. Five to seven stages suits most B2B pipelines; more than that and reps start skipping them.
An ordered set of deal stages representing one distinct sales process.
A pipeline is a named sequence of deal stages. Separate pipelines are justified when the process genuinely differs — new business versus renewals, or direct versus partner-sourced — not when only the product differs. Every extra pipeline multiplies reporting complexity, so the default answer to 'should we add a pipeline?' is no unless the stages themselves would be different.
The staged workflow support tickets move through in Service Hub.
A ticket pipeline defines the statuses a support case moves through, from new to closed, with SLAs and automation attached to each. Separate pipelines make sense when handling processes genuinely differ — technical support versus billing versus onboarding — because each needs different stages, owners and response targets. Reporting on first response and resolution time depends entirely on those stages being applied consistently.
A committed response or resolution time for a defined class of request.
An SLA is a documented commitment: for example, first response within four hours for critical issues and 24 hours for standard requests, measured during business hours. In HubSpot, SLAs are enforced with ticket properties, automation and breach alerts. An SLA without automated measurement is an aspiration — the point is that a missed target creates a visible event, not a private disappointment.
An interactive in-CRM guide for a specific sales conversation.
A playbook is a structured card inside a HubSpot record giving reps the questions, positioning and objection handling for a specific call type, with the answers logged back to properties. It turns discovery from an individual skill into a repeatable process, and it produces structured data about why deals are won and lost — which is usually more valuable than the enablement benefit.
The structured setup of HubSpot from configuration through to team adoption.
HubSpot onboarding is the project of taking a purchased portal to a working system: object and property model, pipelines, automation, integrations, reporting, and role-based training. Standard implementations run four to six weeks; enterprise builds with custom objects and API integrations run eight to twelve. The difference between a good and a bad onboarding shows up in adoption, not in configuration.
02
5 terms
A user-defined record type in HubSpot for entities the standard objects cannot model.
A custom object is a record type you define yourself — Properties, Vehicles, Subscriptions, Courses — with its own properties and associations to contacts, companies, deals and tickets. Available on Enterprise tiers, it is the right answer when an entity has its own lifecycle and needs many-to-many relationships. It is the wrong answer when a property on an existing object, or a well-labelled association, would do. Custom objects add permanent complexity to imports, reporting and integrations.
A name that describes the nature of a relationship between two HubSpot records.
Association labels let you describe why two records are linked: a contact can be the decision maker, billing contact or technical contact on a deal. Without labels, every association looks identical, so reports and workflows cannot distinguish the economic buyer from someone copied on an email. Labels are cheap to set up and are frequently the fix for problems teams try to solve with custom objects.
A container that organises related properties on a record's sidebar.
A property group bundles related properties — for example 'Onboarding' or 'Billing' — so a record page reads as sections rather than a hundred loose fields. Grouping does not affect data, only how usable the CRM feels. Most portals that describe their CRM as cluttered have a grouping and left-sidebar configuration problem rather than a data problem.
A contact you can send marketing email to, and the unit HubSpot bills marketing tiers on.
Marketing contacts are the records HubSpot bills you for in Marketing Hub. Only marketing contacts can receive marketing email or be targeted by ad audiences; non-marketing contacts still exist, keep their history, and can be used in internal automation. Status changes take effect at the start of your next billing period, so cleanup has to be scheduled ahead of renewal rather than after the invoice.
A property whose value HubSpot derives from other properties or associated records.
Calculated properties compute values automatically — sums, differences between dates, minimums and maximums, or rollups from associated records such as total deal value per company. They keep derived numbers consistent instead of relying on workflows to copy values around. Limits apply per tier, and calculations refresh on a schedule rather than instantly, which matters if a workflow depends on the result.
03
10 terms
HubSpot's automation engine: triggers, branches and actions applied to records.
A workflow enrols records that meet trigger criteria and then performs actions — set properties, send internal notifications, create tasks, send marketing email, call a webhook or run custom code. Workflows act on records, run in the background and can be highly branched. They are the right tool for anything that must happen consistently regardless of who is watching, and the wrong tool for personal one-to-one sales outreach.
A one-to-one sales cadence of emails and tasks sent from a rep's own inbox.
Sequences are sales-side cadences: a series of personalised emails and reminder tasks sent from an individual rep's connected inbox, which stop automatically when the prospect replies or books a meeting. Use sequences for individual outreach and workflows for scaled, always-on automation. Sending marketing email through sequences to large lists is the most common misuse and a reliable way to damage domain reputation.
The condition that causes a record to enter a workflow.
An enrolment trigger defines which records enter a workflow and when. Triggers can be property-based, event-based (form submission, page view) or manual. Re-enrolment settings control whether a record can enter again after leaving — powerful for recurring processes and dangerous when combined with a property the workflow itself sets, which is the classic infinite-loop pattern.
A Node.js or Python snippet run inside a HubSpot workflow.
Custom code actions let a workflow execute serverless code with access to HubSpot data and external APIs, returning values the rest of the workflow can branch on. They are available on Operations Hub Professional and above, and they are the right tool for formatting, external lookups and logic HubSpot's UI cannot express. Every custom code action should log its inputs and outputs, because debugging a silent failure inside a workflow is otherwise miserable.
A fixed pattern for naming automations so a portal stays navigable at scale.
A naming convention encodes object, team, action, trigger type and status into every workflow name — for example 'DEAL | SALES | Notify owner on stage change | ACTIVE'. Once a portal passes roughly fifty workflows, the convention is the difference between an operator finding the right automation in seconds and rebuilding a duplicate. Conventions only survive if they are applied retroactively once and enforced at creation.
Software-driven execution of repeatable marketing tasks against defined triggers.
Marketing automation covers the scheduled and triggered execution of marketing activity: nurture email, lifecycle stage updates, list membership, lead routing and internal alerts. Its value is consistency rather than volume — every lead gets the same follow-up regardless of who is on holiday. Its risk is that unmaintained automation keeps running long after the campaign, the person and the reason are gone.
Distributing incoming leads evenly across a group of reps.
Round-robin routing assigns each new qualified lead to the next rep in a rotation, optionally weighted by capacity, territory or specialism. It ensures fairness and speed to lead, which is consistently one of the strongest predictors of conversion. It needs guardrails: skip reps who are out of office, cap daily assignment, and always define what happens when nobody meets the criteria.
An HTTP callback fired when something happens, pushing data to another system.
A webhook sends an HTTP request to a URL you specify when an event occurs, so downstream systems learn about changes immediately instead of polling. In HubSpot, webhooks can be triggered from workflows or subscribed to at the app level. Production webhooks need signature verification, idempotent handling of repeats and a retry strategy, because delivery is at-least-once rather than exactly-once.
Asking for different form fields each time a known contact converts.
Progressive profiling replaces already-known fields on a form with new questions, so a returning visitor is asked something useful instead of retyping their email. It keeps forms short while gradually building a complete record. It requires cookie-based identification to work, so plan for the cases where it fails — new devices, privacy settings and shared machines.
Page or email content that changes based on what is known about the viewer.
Smart content varies what a visitor sees based on lifecycle stage, list membership, country, device or referral source. Used sparingly it improves relevance; used heavily it creates a page nobody can QA and search engines index inconsistently. Keep the default variant complete and self-contained, and treat variants as enhancements rather than as the real content.
04
11 terms
The pattern of pulling data from a source, reshaping it and loading it into a target.
ETL describes the three phases of moving data between systems: extract from the source, transform it to match the destination's model and rules, then load it. In CRM migrations, the transform step carries the risk — field mapping, picklist translation, date formats, deduplication rules and encoding all live there. Anything unmapped at transform time is either lost or dumped into a notes field, which is why the mapping document is signed before any load begins.
A follow-up load that captures records created or changed since the main migration.
A delta sync moves only the records created or modified since the historical snapshot was taken. It is what allows a large migration to run over days while the source system stays in use: the bulk load happens in the background, and a short delta run at cutover catches the difference. Delta syncs depend on reliable last-modified timestamps in the source system — where those are missing, the migration plan needs a different strategy.
The post-migration document proving source and destination counts agree.
A reconciliation report compares source and destination record counts, field fill rates and association counts for every object, and explains every variance. Expected variances (excluded test records, agreed skips) are listed with the rule that produced them; unexplained variances are treated as defects. It is the artefact that lets a business shut down a legacy system with confidence rather than paying for it indefinitely.
Identifying and merging records that represent the same person or company.
Deduplication finds records representing the same entity — matched on email, domain, normalised name or fuzzy similarity — and merges them while preserving associations, activity history and the correct owner. It should be run as its own controlled phase, never simultaneously with a migration, because merging changes counts and makes reconciliation impossible. Every merge should be logged so a mistaken merge can be traced.
The rules and ownership that keep a CRM clean after it has been cleaned.
Data governance is the set of documented rules about who can create properties, what values are permitted, which fields are required at each stage, how imports are approved and who reviews quality. Without it, a cleanup project decays within two quarters. Effective governance is small: a short standards document, validation on the handful of fields that drive routing and reporting, and one named owner.
The signed document defining how every source field lands in the destination.
A field mapping document lists every source field with its destination property, data type, transformation rule and handling for empty or unrecognised values, including an exhaustive enumeration of picklist values. It is the contract for a migration: when someone asks months later why a field looks different, the answer is a row in the document. Unmapped fields are an explicit decision, not an accident discovered later.
The intensive support period immediately after go-live.
Hypercare is the heightened support window straight after launch — typically 30 days — with daily monitoring, rapid fixes, data reconciliation checks and open office hours for questions. It exists because the first two weeks of real use surface issues no test plan predicts. Skipping hypercare is how a technically successful implementation becomes a system nobody trusts.
A separate HubSpot environment for testing changes safely.
A sandbox is a non-production HubSpot portal used to test migrations, automation changes and integrations before they touch live data. Available on Enterprise tiers, it is the difference between discovering a mapping error in a diff report and discovering it in front of the sales team. Sandboxes do not carry all production data by default, so test datasets need deliberate construction.
The cap on how many API requests an integration may make in a window.
HubSpot enforces per-app and per-portal request limits, commonly around 100 requests per ten seconds for standard tiers, plus daily caps. Integrations that ignore them fail silently or partially, leaving data half-synced. Robust integrations batch requests, retry with exponential backoff, queue work during spikes and log every failure so gaps are visible rather than assumed.
Adding third-party firmographic or contact data to existing records.
Enrichment appends data you did not collect — company size, industry, technologies, revenue band, job seniority — from a third-party provider. It shortens forms and improves scoring and segmentation, but coverage and accuracy vary sharply by region and company size. Enrich only the fields that drive a decision, and audit accuracy on a sample before trusting enriched fields in routing.
The scripted checks run in a test portal before changes go live.
A testing plan lists what will be verified before a change ships: record counts, property fill rates, association integrity, workflow outcomes for representative records, integration round-trips and a period of real usage by actual end users. Written as a checklist with named owners, it converts 'we tested it' into evidence someone can review.
05
8 terms
The rule deciding how revenue credit is split across marketing touchpoints.
An attribution model determines which interactions get credit for a conversion: first touch, last touch, linear, U-shaped, W-shaped or time decay. No model is correct — each answers a different question. First touch tells you what creates awareness; last touch tells you what closes. The failure mode is switching models mid-year and treating the resulting change in numbers as a change in performance.
The single system or report designated as authoritative for a given metric.
A source of truth is the one place a number is calculated, which every dashboard and deck then references rather than recreating. Two sources of truth is not redundancy — it is a standing argument, because two independently built reports will diverge on date property, exclusions or window. The fix is a metric dictionary plus one canonical report per metric with a named owner.
A shared document defining every reported metric, its source and its owner.
A metric dictionary records, for each metric: the exact name leadership uses, a one-sentence definition, the object and properties involved, the date property used for filtering, exclusions, the canonical report link and the owner. Twelve to twenty entries covers most mid-market companies. It is the cheapest possible fix for reporting disputes and it takes about a week to write properly.
The function that aligns marketing, sales and service systems, data and process.
Revenue operations is the discipline of running the systems, data and processes behind the whole revenue funnel as one function rather than three. In practice it owns the CRM model, automation, reporting definitions, forecasting hygiene and enablement of the tools. It exists because handoffs between separately optimised teams are where pipeline leaks, and because nobody else is accountable for the end-to-end number.
Senior revenue operations capability retained part-time instead of hired full-time.
Fractional RevOps means retaining an experienced operator for a set number of hours per month rather than hiring a full-time head. It suits companies whose systems need senior judgement regularly but not daily — typically 10 to 40 hours a month covering automation maintenance, reporting, data quality and roadmap. The transition point to in-house is usually when the work becomes continuous rather than project-shaped.
A judgement layer on top of deal stage: commit, best case, pipeline or omitted.
Forecast categories let reps express confidence independently of process stage, so a deal that is technically at proposal stage can still be marked best case rather than commit. They make forecasting more honest, but only if each category has a written definition and a rep is held to their commits. Without definitions, categories become a second, equally vague version of deal stage.
Elapsed time between a lead converting and a rep making first contact.
Speed to lead measures how long it takes for a human to respond after someone raises their hand. It is one of the few operational metrics with a consistently strong relationship to conversion, and it is almost always a routing and notification problem rather than a motivation problem. Measure it from the conversion timestamp to the first logged outbound activity, and report it as a median, not an average.
Reporting from the deal object rather than from contacts or activities.
Deal-based reporting counts revenue events rather than people or touches, which is what leadership usually means when they ask about pipeline. It requires deal records to be created consistently at a defined trigger, amounts to be filled at a defined stage, and close dates to be maintained honestly. Where those three break down, no dashboard configuration will produce a trustworthy forecast.
06
8 terms
A DNS record listing which servers may send email for your domain.
SPF is a TXT record in DNS that names the mail servers authorised to send on behalf of your domain. Receiving servers check the sending IP against it. A domain can have exactly one SPF record, it is limited to ten DNS lookups, and a 'redirect=' modifier is a valid terminal mechanism. Missing or broken SPF is one of the most common causes of legitimate marketing email being filtered.
A cryptographic signature proving an email was authorised by the domain owner.
DKIM adds a signature to outgoing mail headers that receiving servers verify against a public key published in your DNS. It proves the message was authorised by the domain and has not been altered in transit. HubSpot provides the records to publish when you connect a sending domain; without DKIM, marketing email is far more likely to be filtered and DMARC cannot pass.
A policy telling receivers what to do when SPF and DKIM checks fail.
DMARC builds on SPF and DKIM by publishing a policy — none, quarantine or reject — plus a reporting address. It also enforces alignment between the visible From domain and the authenticated domain, which is what actually stops spoofing. Best practice is to start at p=none with reporting, read the reports for a few weeks, then tighten to quarantine and finally reject.
A mailbox provider's running assessment of whether your mail is wanted.
Domain reputation is the score mailbox providers assign to your sending domain based on engagement, complaint rates, bounce rates and spam trap hits. It moves slowly downward and even more slowly upward. The practical levers are list hygiene, suppressing unengaged contacts, honouring unsubscribes immediately and avoiding sudden volume spikes on a cold domain.
An address used to identify senders with poor list hygiene.
Spam traps are addresses that never opt in — either created purely to catch spam, or recycled from long-abandoned mailboxes. Hitting one signals that you are mailing purchased, scraped or very stale lists, and the reputation damage is disproportionate. The defence is never buying lists, using double opt-in for cold sources, and suppressing contacts with no engagement for twelve months.
Routinely removing or suppressing addresses that harm deliverability.
List hygiene is the ongoing practice of suppressing hard bounces immediately, removing role addresses where appropriate, re-permissioning or suppressing contacts with no engagement in twelve months, and validating addresses collected from offline sources. It reduces cost as well as risk: in HubSpot, unengaged contacts left in marketing status are billable as well as harmful.
A list of addresses that must never receive marketing email.
A suppression list holds unsubscribes, hard bounces, complainers and any address you are legally or commercially required not to mail. It is checked before every send and takes precedence over every other list. Maintaining suppression properly is both a compliance requirement and a deliverability lever, because repeatedly mailing complainers damages reputation quickly.
Requiring email confirmation before a subscriber is added to a mailing list.
Double opt-in sends a confirmation email after signup and only subscribes the address once the link is clicked. It reduces list size and increases list quality: fewer typos, no spam traps from mistyped addresses, and demonstrable consent for GDPR purposes. It is strongly recommended for cold or high-volume acquisition sources and often unnecessary for demo requests.
07
7 terms
Structuring content so AI assistants retrieve and cite it directly.
Answer engine optimisation is the practice of writing and structuring content so assistants like ChatGPT, Perplexity and Google AI Overviews can extract a complete answer from a single passage. In practice it means question-shaped headings, self-contained 50–150 word answer paragraphs, tables and numbered procedures, accurate schema, and content present in server-rendered HTML. The unit of optimisation is the passage, not the page.
Influencing how generative AI systems describe your brand and category.
Generative engine optimisation is the broader discipline of shaping how AI systems represent you: which facts they associate with your brand, which competitors they name alongside you, and which sources they draw on. Beyond your own site it depends on corroboration — consistent descriptions across directories, review platforms, solution provider listings and press — because models weight claims that appear across independent sources.
Splitting content into self-contained passages that survive retrieval.
Chunking is how retrieval systems divide documents into passages before embedding and indexing them. A chunk that depends on earlier context — pronouns, 'as mentioned above', a definition three paragraphs up — becomes incoherent once separated, so it is unlikely to be retrieved or cited. Writing chunk-ready means every section under a heading answers its own question completely.
Grounding an AI model's answer in documents retrieved at query time.
Retrieval-augmented generation combines a language model with a search step: the system retrieves relevant passages from a document store and passes them to the model as context for the answer. It is how AI assistants cite live sources and how internal AI assistants answer questions over company data. Answer quality depends far more on retrieval and chunking quality than on which model is used.
Structured data in JSON-LD that tells machines what a page is about.
Schema markup is machine-readable structured data, usually embedded as JSON-LD, describing the entities on a page: Organization, Service, Article, FAQPage, DefinedTerm, BreadcrumbList. It removes ambiguity for search engines and AI systems about what you are, what you offer and how questions map to answers. Schema must describe what is visibly on the page — mismatched markup is worse than none.
A short self-contained paragraph that fully answers the heading above it.
An answer block is a 50–150 word paragraph placed immediately under a question-shaped heading that answers the question completely, with no dependency on surrounding text. It is the unit AI systems retrieve and quote. The test is simple: copy it into a blank document and check whether it still answers the question for someone who has never seen the site.
Generating complete page HTML on the server before it reaches the client.
Server-side rendering means the HTML sent in the initial response already contains the page content, rather than an empty shell that JavaScript fills in. It matters for AI visibility because most AI crawlers do not execute JavaScript — content that only appears after hydration is invisible to them. The check takes seconds: request the page and read the raw HTML.
Next step
If several of these definitions describe problems you currently have, a 30-minute call will turn them into a sequence of fixes with costs attached.