Glossary

B2B data & cold email, defined.

61 terms from contact data, email verification, deliverability, and outbound automation — each defined precisely enough to act on. For the verification methodology itself, read how we verify.

Account-based marketing (ABM)

A go-to-market motion that inverts lead generation: instead of casting wide and qualifying inbound, you pick a finite list of named target accounts and concentrate marketing and sales effort on them in coordination. ABM suits high-ACV products with small TAMs, where any individual account justifies bespoke effort. Operationally it needs three data layers: a clean account list matching the ICP, mapped buying committees with verified contact channels inside each account, and timing signals to prioritize. The common failure is doing ABM-grade targeting with list-blast execution — the account selection is strategic but the touch is generic.

Agent workflow

A multi-step task executed autonomously by an AI agent calling tools in sequence: for outbound, something like "pull companies matching the ICP, find the buying committee at each, verify emails, export valid contacts, draft openers, hand off to the sequencer." Workflows differ from scripts in that the agent decides branching at runtime — retrying a failed lookup, skipping an account with no fit. Reliability engineering for agent workflows centers on the tool layer: idempotent operations, clear error semantics, rate limits the agent can respect, and spend controls so a looping agent cannot drain a credit balance. MCP is the emerging standard interface for these tool calls.

AI SDR

Software that performs the work of a sales development rep with an LLM in the loop: selecting accounts, finding and verifying contacts, writing personalized first-touch emails, sequencing follow-ups, and triaging replies. Current systems are strongest at research and drafting, weakest at judgment — knowing when not to send, and handling nuanced replies. The data layer constrains everything: an AI SDR with stale or unverified contacts automates bounces at machine speed, damaging domains faster than a human ever could. Practical deployments keep a human review gate on outgoing copy and feed the agent verified, per-valid-priced data through an API or MCP connection.

API key

A secret token that identifies and authorizes a client calling an API; in data products it also scopes billing, rate limits, and permissions. Keys are bearer credentials — anyone holding one is you — so the operational rules are strict: store them in environment variables or a secrets manager, never in client-side code or repositories; issue separate keys per environment and integration so one leak doesn't force rotating everything; rotate on any suspicion of exposure. Argorant issues API keys on paid plans for both the REST API and the MCP server, alongside a command-line client for scripted access.

B2B contact database

A searchable repository of companies and the professionals working at them, combining firmographics, role and seniority data, and contact channels such as business email addresses. Quality varies on three axes: coverage (how much of your TAM is present), accuracy (whether the data reflects current reality), and verifiability (whether emails are checked before you pay for them). Argorant's database covers 614M contacts, lets you filter and count for free, and verifies every email by live SMTP probe at export — invalid addresses cost zero credits. The economics of a database matter as much as its size: paying for bounces is paying twice.

Bounce rate

The percentage of sent messages that fail to deliver, counting both hard and soft bounces. Mailbox providers and sending platforms watch it closely: sustained bounce rates above roughly 2% suggest list-quality problems, and rates above 5% commonly trigger throttling, spam-folder placement, or account suspension by the sending platform itself. Because hard bounces dominate the damage, the most effective lever is removing nonexistent mailboxes before sending — which is what pre-send SMTP verification does. Bounce rate is measured per sending domain and per campaign, so one bad list can poison a domain shared across many sequences.

Buying committee

The group of people inside an account who collectively decide a B2B purchase: the economic buyer who controls the spend, the champion who pushes internally, technical evaluators, end users, procurement, and often security or legal. Mid-market and enterprise deals routinely involve five to ten such people. For outbound this argues against single-threading — sequencing one contact per account leaves the deal hostage to one inbox. Multi-threading means identifying several committee roles per account up front and tailoring the message to each: ROI for the buyer, workflow relief for the user, integration answers for the evaluator.

Catch-all domain

A domain whose mail server accepts mail for every address at that domain, whether or not the mailbox exists. Because the server answers "yes" to any RCPT TO probe, SMTP verification cannot distinguish a real mailbox from a fabricated one. No vendor can verify through a genuine catch-all; claims of definitive per-mailbox verdicts on these domains are guesses. The honest treatment is to flag the address as catch-all and let the sender decide whether the residual bounce risk is acceptable. Argorant marks catch-alls explicitly and excludes them from exports unless you opt in.

Catch-all flag

The status a verifier assigns when a domain accepts mail for any address, making a per-mailbox verdict impossible. The flag is an honest "cannot know," distinct from both valid and invalid, and how a vendor handles it reveals their integrity: silently counting catch-alls as valid inflates apparent accuracy and bills you for unverifiable addresses. Argorant flags catch-alls explicitly, leaves them out of exports by default, and includes them only when you opt in — at which point they cost a credit like any other delivered contact.

Cold email

Unsolicited one-to-one business email sent to a prospect with whom the sender has no prior relationship, typically to start a sales conversation. It differs from marketing email in mechanics and law: it is sent from individual inboxes through normal mail infrastructure rather than bulk ESPs, personalized per recipient, and in most jurisdictions governed by B2B-specific rules (CAN-SPAM in the US; legitimate-interest analysis under GDPR in the EU) rather than newsletter consent rules. Effective cold email depends on tight ICP targeting, verified addresses, warmed domains, low per-inbox volume, and a working opt-out path.

CRM sync

Keeping contact and account records consistent between a data source and a CRM such as HubSpot, Salesforce, or Pipedrive. Sync covers initial import, field mapping (whose "company size" wins), update propagation, and conflict rules when both sides changed. The recurring failure modes are duplicate creation from weak matching keys, silent overwrites of rep-edited fields, and one-way syncs masquerading as two-way. A defensible setup deduplicates on email and domain before writing, marks provenance on enriched fields, and never overwrites human-entered data without an explicit rule. CSV remains the lowest-friction sync mechanism, and the easiest to audit.

CSV export

Downloading filtered, verified contacts as a comma-separated file — still the universal interchange format between data sources, sequencers, and CRMs because every tool on both sides can read it. The details that bite in practice: UTF-8 encoding for non-ASCII names, fields quoted against embedded commas, one header row with stable column names, and a verification-status column so downstream tools can act on catch-all flags. On Argorant, the export step is where verification happens — each address is SMTP-probed as the file is built, and only valid (plus explicitly opted-in catch-all) rows consume credits.

Data decay

The steady invalidation of contact data as reality moves: people change jobs, companies rename, merge, and shut down, mail systems migrate. B2B contact data is commonly estimated to decay around 25–30% per year, with email addresses among the fastest-decaying fields because they break the day someone leaves. Decay is why a database's collection date matters more than its size, and why verification must happen close to the moment of use: an address checked six months ago carries meaningful bounce risk today. Freshness windows and scheduled re-verification are the operational responses.

Data enrichment

Filling in missing attributes on records you already have: appending a verified email to a name-and-company pair, adding firmographics to a domain, or attaching seniority and department to a job title. Enrichment is matching plus lookup — the input record is matched against a reference database, and missing fields are copied over. Its failure modes are mismatches (wrong person at a same-named company) and stale appends. Enrichment runs as one-off list processing, as continuous CRM hygiene, or as a waterfall across multiple steps. The unit economics question is always the same: do you pay per attempt or per successful, verified append.

Dedupe (deduplication)

Removing duplicate records that refer to the same person or company. Exact-match deduping on email address is trivial; the hard cases are fuzzy — the same person under j.smith@ and john.smith@, the same company as "Acme Inc" and "Acme Incorporated", or one human present in both an imported list and a CRM. Practical dedupe keys on normalized email first, then person-plus-domain, then probabilistic name matching. In outbound the stakes are concrete: a duplicate means the same prospect receives two parallel sequences, which reads as carelessness and doubles complaint risk. Dedupe against your CRM and suppression list before every import.

Deliverability rate

The share of sent messages that reach recipients' inboxes — not merely accepted by the server, but placed in the inbox rather than spam. It is harder to measure than bounce rate because spam-folder placement generates no error: the message simply goes unseen. Teams estimate it through seed-list tests, placement-monitoring tools, and indirectly through open-rate collapse across a whole campaign. Drivers are sender reputation, authentication, content, volume pacing, and list quality. A list can have a 0% bounce rate and still suffer poor deliverability if the sending domain's reputation is damaged.

Disposable email

A temporary address from services that provide throwaway inboxes, typically created to pass a signup form and abandoned minutes later. Disposable domains are catalogued in public and commercial blocklists, and competent verification flags them as a distinct status: the mailbox may technically accept mail at probe time, yet sending to it is pointless and signals a low-quality list. In B2B data the related problem is personal-domain noise — freemail addresses where a business mailbox is expected. Both classes belong in list hygiene filters, removed before a campaign rather than discovered through silence afterward.

DKIM (DomainKeys Identified Mail)

A cryptographic signature added to outgoing mail headers, verifiable against a public key published in the sending domain's DNS. DKIM proves the message was authorized by the domain and was not altered in transit, and unlike SPF it survives forwarding. The receiving server fetches the public key from the selector record (e.g. selector._domainkey.example.com) and validates the signature. For cold-email infrastructure, every sending domain needs its own DKIM key configured at the email provider; a missing or misaligned DKIM signature causes DMARC failures and sharply worse inbox placement at major mailbox providers.

DMARC

A DNS policy that tells receiving servers what to do when a message fails SPF and DKIM alignment: do nothing (p=none), quarantine to spam, or reject. Alignment means the authenticated domain matches the visible From domain, closing the gap both underlying protocols leave open. DMARC also enables aggregate reports showing who is sending as your domain. Since 2024, major mailbox providers require DMARC from bulk senders, making it effectively mandatory for outbound. Sensible practice: start at p=none to observe, then move to quarantine or reject once legitimate mail streams are aligned.

Domain warm-up

The practice of gradually building sending history on a new domain or inbox before using it for real outbound. Mailbox providers distrust domains with no track record, so an immediate burst of cold email from a fresh domain is filtered aggressively. Warm-up ramps volume over several weeks — often starting at 5–10 messages per day per inbox — while generating positive engagement signals. Many teams use dedicated secondary domains for cold email so the primary corporate domain's reputation is never at risk, warming each one before it enters rotation in a sequencer.

Email deliverability

The discipline of getting email into the recipient's inbox rather than the spam folder or a rejection. It spans authentication (SPF, DKIM, DMARC), sender reputation, list quality, sending infrastructure, volume pacing, and content. Deliverability differs from delivery: a message can be accepted by the server (delivered) yet filed as spam (not deliverable in the practical sense). For cold email specifically, the controllable inputs are verified recipient addresses, warmed sending domains, conservative daily volumes per inbox, and correct authentication records. Everything downstream — open rate, reply rate, pipeline — depends on deliverability first.

Email finder

A tool that returns a person's business email address given their name and company or domain. Under the hood it combines database lookup (the address has been observed before) with pattern inference (the company uses first.last@domain, so generate that and verify it). The verification step is what separates a finder from a guesser: an inferred address is a hypothesis until an SMTP probe confirms the mailbox exists. Reasonable pricing charges only for found-and-valid results, since an unverified guess shifts the bounce risk — and the cost — onto the sender.

Email pattern

The convention a company uses to form employee addresses: first.last@, flast@, first@, or rarer variants. Patterns are derived statistically — observe enough confirmed addresses at a domain and the dominant format emerges. Knowing the pattern lets a finder construct a candidate address for any employee, but a pattern is probabilistic, not a guarantee: companies change conventions after rebrands, large firms run several patterns concurrently, and collisions (two J. Smiths) force exceptions. A pattern-derived address therefore needs individual SMTP verification before it is worth sending to, and on catch-all domains the pattern can never be confirmed mailbox-by-mailbox.

Email syntax check

The first and lightest layer of verification: does the string conform to the structure of an email address — valid local part, exactly one @, a resolvable-looking domain with a plausible TLD. Syntax checks also catch mechanical junk: whitespace, typo'd domains (gamil.com), and commas where dots belong. Passing syntax says nothing about whether a mailbox exists; it only earns the address the next checks — domain and MX lookup, then a live SMTP probe. In a verification pipeline, syntax failures are rejected instantly and never reach the network, which is why they are free almost everywhere.

Firmographics

Attributes that describe a company rather than a person: industry, employee count, revenue band, geography, founding year, legal form, and growth stage. They are the demographic variables of B2B and the backbone of ICP filters and territory design. Firmographics are relatively stable but not static — headcount bands shift, companies relocate and get acquired — so a database's value depends on how recently they were observed. They combine with technographics (what the company runs) and intent data (what it is researching) to move from "could buy" to "likely to buy now."

Freshness window

The period during which a verification result is treated as still trustworthy, so the same address is not re-probed on every use. Mailbox state changes with job moves and mail-system migrations, but not minute to minute. Argorant sidesteps the staleness question entirely by verifying at the moment of export: every address is probed live at download, so the result reflects current mailbox state rather than a stored timestamp.

Greylisting

An anti-spam technique where a receiving server temporarily rejects mail from an unknown sender with a 4xx code, expecting a legitimate server to retry after a delay — which most spam software historically did not. For delivery, greylisting just adds latency. For verification, it is a complication: the probe receives a temporary rejection that says nothing about whether the mailbox exists. A careful verifier retries after the greylisting window or returns an honest "unknown" rather than coercing the result into valid or invalid. Treating unknowns as invalid — undelivered and unpaid — is the conservative handling.

Hard bounce

A permanent delivery failure: the receiving server rejects the message outright, typically with a 5xx code, because the mailbox does not exist, the domain has no mail service, or the recipient has blocked the sender. Hard bounces are the most damaging event in outbound email because mailbox providers read them as evidence that the sender does not know their recipients — a hallmark of purchased or stale lists. Repeated hard bounces degrade sender reputation quickly and can trigger blocklisting. Verifying addresses by SMTP before sending removes nearly all hard bounces, since the same rejection the bounce would produce surfaces during the probe instead.

ICP (Ideal Customer Profile)

A precise description of the company most likely to buy, retain, and expand — defined by firmographics (industry, headcount, revenue, geography), technographics, and situational signals like hiring or funding. The ICP describes companies; buyer personas describe the people inside them. In outbound, the ICP translates directly into database filters: it determines which of millions of companies are worth a credit and which are noise. A vague ICP produces large, low-converting lists; a sharp one produces smaller lists with higher reply rates and less reputational risk per send. Revisit it whenever closed-won data contradicts it.

Intent data

Behavioral signals suggesting a company is actively researching a problem or product category: content consumption on publisher networks (third-party intent), visits to your own site (first-party), hiring for relevant roles, new funding, or technology changes. Intent data answers "who is in-market now" where firmographics answer "who fits." Its weaknesses are noise and latency — surges can come from a single curious employee, and signals are aggregated at company level, not person level. Used well, intent reorders an already-qualified ICP list by urgency rather than expanding it with unqualified accounts.

Lead routing

The rules that decide which rep or queue receives a new lead: by territory, company size, industry, product line, round-robin, or account ownership. Routing quality depends directly on data quality — a lead with no reliable country or headcount field cannot be routed by territory or segment, so it lands in a default queue and goes stale. Speed matters too; response within minutes rather than hours measurably improves conversion on inbound. For outbound-sourced leads, routing usually keys on existing account ownership first to avoid two reps colliding on one company.

Lead scoring

Ranking leads by expected value using fit and behavior. Fit scoring grades how closely the company and contact match the ICP — industry, size, seniority — and can be computed the moment a record enters the system. Behavioral scoring adds engagement: site visits, content, product signups. Scores route attention: high-fit-high-intent goes to a rep today, high-fit-low-intent enters nurture, low-fit is declined regardless of activity. Scoring quality is bounded by data quality; a model fed empty or stale firmographic fields ranks on noise. Simple transparent rules usually beat opaque models early, because reps must trust the score to act on it.

List hygiene

The ongoing practice of keeping a contact list sendable: verifying addresses before use, removing hard bounces immediately, suppressing opt-outs and complaints, filtering role-based and disposable addresses, deduplicating, and re-verifying anything that has aged past its freshness window. Hygiene costs less than its absence — every bad address removed before sending is a bounce, complaint, or trap hit that never reaches your domain's reputation. Mature teams treat hygiene as a pipeline stage, not an occasional cleanup: data enters verified, decays on a known clock, and is re-checked or retired on schedule.

List verification

Running an existing email list — uploaded from a CRM, a previous vendor, or an old campaign — through a verification pipeline, returning each address with a status: valid, invalid, catch-all, unknown, plus hygiene flags like role-based or disposable. It differs from verified sourcing: here the addresses already exist and the job is to find out which still work. The output drives a split decision — send to valid, drop invalid, quarantine catch-alls and unknowns. On Argorant, verifying an uploaded list costs one credit per check. Any list older than a couple of months should be verified before it touches a warmed domain.

Masked preview

Showing a contact record with the sensitive field partially hidden — j*****@acme.com — so a user can judge relevance before spending a credit to reveal it. Masking solves a trust asymmetry in data products: buyers want proof the data exists and matches their filters; sellers cannot expose the asset for free. A useful masked preview keeps enough signal to evaluate (domain visible, name and title in full, verification status shown) while withholding exactly the deliverable part. On Argorant, searching and counting masked results is free; the credit is spent only when a contact is exported and its address verifies as valid.

MCP (Model Context Protocol)

An open protocol that lets AI assistants and agents call external tools and data sources in a standardized way. An MCP server exposes typed tools — search contacts, verify an email, export a list — and any MCP-capable client (Claude, IDE agents, custom harnesses) can discover and invoke them without bespoke integration code. For B2B data this collapses the distance between question and answer: an agent can be asked for "verified CTOs at German manufacturers over 200 employees" and execute the search, verification, and export itself. Argorant ships an MCP server alongside its REST API on every paid plan.

MX record

The DNS record that names the mail servers responsible for receiving email for a domain, with priority values controlling failover order. MX records are the first stop in both delivery and verification: no MX (and no fallback A record) means the domain cannot receive mail at all, which invalidates every address on it without a single SMTP probe. The MX hostname also reveals the mail provider — Google Workspace, Microsoft 365, or self-hosted — which predicts how the server will respond to probes and how aggressive its filtering is likely to be.

OAuth

An authorization framework that lets a user grant one application limited access to their account on another, without sharing a password. The user is redirected to the provider, approves a scoped consent screen, and the application receives short-lived access tokens plus a refresh token. OAuth differs from API keys in audience and lifecycle: keys suit server-to-server machine access under your own account, while OAuth suits acting on behalf of a user — connecting a CRM, or letting an AI client authenticate to an MCP server. Scopes matter: requesting more access than a feature needs is both a security smell and a review obstacle on app marketplaces.

Open rate

The percentage of delivered messages registered as opened, measured by a tracking pixel loading in the recipient's client. Once the headline metric of email, it is now structurally unreliable: Apple Mail Privacy Protection pre-fetches pixels for a large share of recipients, inflating opens, while image-blocking suppresses real ones. Many cold-email teams disable open tracking entirely, because the tracking domain and pixel are themselves spam-filter signals. Open rate retains one honest use: a sudden collapse across a campaign usually means deliverability broke, not that subject lines got worse. For outcome measurement, reply rate is the sturdier metric.

Opt-out

A recipient's request to stop receiving messages, which must be easy to make and honored promptly — under CAN-SPAM within 10 business days, under GDPR without undue delay, and in practice immediately, because continuing to mail someone who objected converts irritation into complaints and legal exposure. In cold email the mechanism is usually a plain-text line offering to stop on reply, or an unsubscribe link; either must actually work and must propagate to the suppression list across all tools and future list imports. An opt-out also extinguishes the legitimate-interest basis for that person's data in your outbound system.

Outbound agency

A service firm that runs cold outbound for clients end to end: ICP definition, list building, domain and inbox infrastructure, copywriting, sequencing, and reply handling, typically billed as retainer plus performance. Agencies are heavy data consumers — each client needs fresh, verified lists in a distinct ICP — so their economics are sensitive to per-contact cost and bounce-driven domain burn. Evaluating one: ask who owns the domains and data when the engagement ends, how they verify lists before sending, and whether reporting is on replies and meetings rather than volume sent. For agencies, pay-per-valid data pricing maps cleanly onto per-client cost accounting.

Pay-per-valid

A pricing model where you are charged only for addresses that verify as deliverable; invalid, unknown, and (unless you opt in) catch-all results cost nothing. It matters because it aligns incentives: a vendor charging per record profits from padding lists with stale data, while a vendor charging per valid result eats the cost of its own decay. Argorant prices this way — every export is SMTP-verified and invalid addresses cost zero credits. The model also changes how you evaluate database size: raw contact counts matter less when you only ever pay for the verified subset.

Rate limit

A cap on how many API requests a client may make per unit of time, enforced to protect shared infrastructure and meter usage fairly. Limits are typically expressed per key per minute, signalled with HTTP 429 responses, and accompanied by headers indicating remaining quota and reset time. Well-behaved clients — and well-built agents — implement exponential backoff with jitter rather than hammering retries, and batch operations where the API allows it. In data APIs, rate limits also interact with billing: the limit governs request velocity while credits govern consumption, and hitting one is not the same as exhausting the other.

Related: API key, Webhook, Agent workflow

Re-verification

Checking an address again after time has passed since its last result. Because data decays continuously, a verification is a timestamped fact, not a permanent property: "valid in January" is a weak claim in June. Re-verification policy is where vendors diverge — some verify once at ingestion and never again, leaving the decay risk with you; Argorant re-probes automatically at export, so the result you pay for reflects current mailbox state. For your own stored lists, the equivalent discipline is re-verifying before any campaign that follows a multi-month gap.

Reply rate

The percentage of delivered messages that receive a reply — the cleanest outcome metric in cold email, immune to pixel-tracking distortions. It compounds three factors: deliverability (did it arrive), targeting (was this the right person), and message quality (was it worth answering). Benchmark expectations vary by market, but sustained rates under ~1% usually indicate a deliverability or ICP problem rather than copywriting. Raw reply rate counts negative replies too; positive-reply rate is the stricter cut. Because replies also signal engagement to mailbox providers, a replied-to thread actively improves the sending domain's standing.

Right to be forgotten

The GDPR right to erasure (Article 17): a person can require a controller to delete their personal data when it is no longer needed, when they object to legitimate-interest processing, or when it was unlawfully processed. For outbound operations this means a deletion request must reach every store holding the record — the database vendor, the CRM, the sequencer, exported CSVs — not just the tool the request arrived in. A subtlety: to honor "never contact me again," most controllers retain a minimal suppression entry (typically a hashed address), since full deletion would make re-import undetectable. Erasure and suppression are complements, not contradictions.

Role-based email

An address tied to a function rather than a person: info@, sales@, support@, hr@. These mailboxes are often read by several people or routed into ticketing systems, and they generate disproportionate spam complaints because no individual opted into anything sent there. Many sending platforms and verification tools flag role accounts separately, and most cold-email practitioners exclude them: a sequence written for a person reads badly in a shared inbox. The legitimate use is structural outreach — partnership or press inquiries — where the function, not the individual, is genuinely the addressee.

Sender reputation

The score mailbox providers assign to a sending domain and IP, built from bounce rates, spam complaints, spam-trap hits, engagement signals, and authentication status. It determines whether your mail lands in the inbox, the spam folder, or is rejected outright. Reputation is earned slowly through consistent low-bounce, low-complaint sending and lost quickly through a single bad blast. It attaches to the domain, so damage outlives any individual campaign. The standard protections are SPF/DKIM/DMARC alignment, gradual warm-up of new domains, verified recipient lists, and honoring suppression lists immediately.

Seniority

A normalized level attached to a contact's role — commonly owner, C-suite, VP, director, manager, senior IC, IC — inferred from job titles. Normalization is the hard part: titles are inconsistent across companies, languages, and inflation cycles (a "VP" at a bank is mid-level; a "Head of" at a startup may be the entire department). Seniority filters are how outbound targets decision-makers versus practitioners, and the right level depends on deal size: small deals close with managers, large ones need the economic buyer plus the people who will use the product.

Sequencer

Software that sends multi-step cold-email campaigns automatically: an initial message followed by timed follow-ups that stop when the prospect replies, bounces, or opts out. Modern sequencers rotate sending across many inboxes and domains to keep per-inbox volume low, randomize send times, track opens and replies, and manage suppression lists. The sequencer is the consumer of your contact data — its results are bounded by list quality, which is why verified exports matter: a sequencer faithfully sending to dead mailboxes will burn the domains it rotates through. Common examples are tools in the Smartlead/Instantly/Lemlist category.

SMTP verification

A live check that asks the receiving mail server whether a specific mailbox exists, without sending an email. The verifier connects to the server listed in the domain's MX record, issues the SMTP commands a real sender would (HELO, MAIL FROM, RCPT TO), reads the response code, and disconnects before any message is transmitted. A 250 response means the mailbox is accepted; a 550 means it does not exist. It is the strongest signal available for mailbox validity, though catch-all domains and greylisting servers can blunt it. Argorant runs SMTP verification at export time, so the result reflects the mailbox's state when you actually receive the address.

Soft bounce

A temporary delivery failure, usually signalled with a 4xx SMTP code: the mailbox is full, the server is briefly unavailable, the message is too large, or the receiver is greylisting the sender. Most sending platforms retry soft bounces for a period before giving up. Unlike hard bounces, soft bounces do not necessarily indicate a bad address, so they carry less reputational weight — but a mailbox that soft-bounces persistently is often abandoned and should be treated as undeliverable. Verification cannot eliminate soft bounces, because they reflect transient server state rather than mailbox existence.

Spam trap

An email address operated by a blocklist or mailbox provider purely to catch senders with bad list practices. Pristine traps are addresses that never belonged to a real person, seeded where scrapers harvest; sending to one proves the list was scraped. Recycled traps are abandoned mailboxes that bounced for a period and were then reactivated as traps; hitting one proves the list is stale and unverified. Trap hits damage sender reputation far more than ordinary bounces and can land a domain or IP on public blocklists. Recent verification is the practical defense — recycled traps bounce during their dormant phase, so a probe catches them before reactivation works against you.

SPF (Sender Policy Framework)

A DNS TXT record listing which servers are authorized to send email on behalf of a domain. When a message arrives, the receiving server checks the connecting IP against the SPF record of the envelope-from domain; a failure marks the message as potentially forged. SPF alone is weak — it breaks on forwarding and says nothing about the visible From header — which is why it is paired with DKIM and enforced through DMARC. An SPF record has a 10-DNS-lookup limit; exceeding it through nested includes silently invalidates the record, a common and easy-to-miss configuration error.

Suppression list

A do-not-contact list checked before every send: people who opted out, addresses that hard-bounced, domains of existing customers or open deals, competitors, and anyone whose data you are legally required to stop processing. Suppression is subtractive and absolute — a match removes the contact regardless of how well they fit the ICP. It must be enforced at every layer that can send: the sequencer, the CRM, and any agent acting through an API, because an opt-out honored in one tool and missed in another is still a violation. Suppression lists should survive tool migrations; they are among the few datasets in outbound that only ever grow.

TAM (Total Addressable Market)

The full set of companies that could plausibly buy your product, usually expressed as a count and a revenue ceiling. In outbound practice, TAM is operational rather than theoretical: it is the number of accounts your ICP filters return from a contact database, which dictates how long outbound can run before you exhaust the market and must widen targeting or wait for re-engagement windows. Counting TAM should be free and repeatable — you want to test filter combinations without spending anything — and the count is also a sanity check on whether your ICP is too narrow to sustain pipeline.

Technographics

Data about the technology a company uses: its CRM, cloud provider, ecommerce platform, analytics stack, programming frameworks, and security tooling. Detected from website source, DNS records, job postings, and integration marketplaces. Technographics sharpen targeting in two ways: as qualification (sell a Shopify app only to Shopify stores) and as displacement signals (target users of the competitor you replace). They decay faster than firmographics because stacks change with every replatforming, so detection recency matters. In messaging, a verified technographic is one of the few personalization hooks that is specific without being invasive.

Verification credit

The billing unit for email checks. One credit buys one verification event, though what counts as a billable event varies by vendor: some charge per attempt regardless of outcome, others only for definitive results. On Argorant, exporting a contact consumes a credit only when the address verifies as valid, and verifying a list you upload costs one credit per check. Credits make costs proportional to usage rather than seat count, which suits the spiky cadence of outbound — heavy spend while building campaigns, near zero between them. When comparing vendors, the question is what a credit actually guarantees: an attempt, a verdict, or a deliverable address.

Waterfall enrichment

Running an enrichment request through multiple steps in sequence, falling through to the next only when the previous one fails to return a usable result. The order is usually set by accuracy and cost: try the strongest match first, fall back to broader inference, and verify whatever comes back before accepting it. Waterfalls raise aggregate match rates well above any single step, at the price of latency and more complex billing — which is why per-valid pricing pairs naturally with them: you pay once for the confirmed result, not for every step that tried.

Webhook

An HTTP callback: instead of your system polling an API for status, the API calls a URL you register when an event occurs — a list finishes verifying, an export completes, a credit threshold is crossed. Webhooks suit asynchronous jobs like bulk verification, where polling wastes rate limit and adds latency. Production-grade handling means verifying the payload signature so forged calls are rejected, responding quickly and processing asynchronously, and treating deliveries as at-least-once — the same event may arrive twice, so handlers must be idempotent. The receiving endpoint becomes part of your security surface and deserves the same care as any public route.

Put the vocabulary to work.

Free account, 100 credits included. Counts and searches are always free.

Get started Connect your AI agent