Glossary

The vocabulary, defined honestly

Every term you'll hit in GTM operations and engineering, in plain language with the practitioner's take. Not a textbook: what the word means, and what it means when it goes wrong.

Pipeline & forecasting

Pipeline coverage: Open pipeline divided by the target it has to produce. The honest number is not a flat 3x, it is 1 divided by your win rate: at a 20% win rate you need 5x, not 3x.
Weighted pipeline: Each deal multiplied by its stage probability. Useful for a roll-up, dangerous as a forecast: a $500K deal at 20% forecasts the same as five $100K deals at 20%, and they behave nothing alike.
Commit: The deals a rep is willing to bet on this period. A commit is only real if the close date is in-period and you can name what has to be true for it to land.
Slippage: A committed deal that pushes to a later period. Two close-date pushes on a commit deal is a tell, not a coincidence.
Stage exit criteria: The observable facts required to advance a deal: economic buyer confirmed, next meeting booked, decision process documented. If the criteria are vibes, weighted pipeline is fiction.
Forecast accuracy / variance: How close the called number lands to actuals. Elite teams hold inside 5% variance; most miss by 10%+, usually from missing cadence, not a bad model.
Best case: The optimistic category above commit: deals that land only if everything breaks right. If your best case and your commit are the same number, one of the two categories is theater.
Pipeline generation / pipegen: Net-new qualified pipeline created in a period. It lags bookings by a full sales cycle, so a pipegen miss this quarter is a bookings miss two quarters out.
Win rate: Deals won over deals closed (won plus lost). Report it by stage entered, not by created date, or a rep who disqualifies fast looks worse than one who lets zombies rot.
Sales cycle length: Calendar days from opp created to closed won. Use the median, never the mean: three 400-day whales drag the average and hide that most deals close in 60.
Stage conversion rate: Share of deals that move from one stage to the next. The stage with the worst conversion is your real bottleneck, not the one holding the most deals.
Roll-up forecast: Summing rep commits up through managers to one number. Every layer adds a haircut or a prayer; with no category discipline it drifts 20%+ from actuals.
Pull-forward: Closing a later-period deal early to cover a current gap. Feels like a win, robs next quarter; two straight quarters of it is a pipeline problem in disguise.
Pipeline snapshot / historical pipeline: A point-in-time capture of the pipeline as it looked then, not as rewritten since. Live CRM lies about the past, so without snapshots you cannot measure slippage honestly.
Deal desk: The cross-functional gate (RevOps, finance, legal) that reviews nonstandard deals before they close. Its job is to say yes fast; if it adds three days to every deal it is misconfigured.
Linearity: How evenly bookings land across the quarter versus stacking in the last two weeks. Bad linearity is not a closing style, it is a slippage and qualification problem you are calling culture.

Metrics

NRR / net revenue retention: Starting ARR plus expansion minus churn and contraction, over starting ARR. Above 100% the book grows on its own; below 100% new sales refill a leak before they add anything. Median is ~102%, best-in-class 120%+.
GRR / gross revenue retention: Retention before any expansion is counted. Caps at 100%. The honest floor metric; NRR can hide churn that GRR exposes. Median ~84%.
CAC payback: Months to recover the cost of acquiring a customer. Under 12 is efficient, 12 to 18 is normal, past 24 you are buying revenue faster than it pays you back. PLG runs 6 to 9; enterprise up to 24.
Magic number: Net-new ARR over prior-period sales and marketing spend. Above 0.75 you should spend more; below 0.5 you should fix the funnel before adding budget.
Pipeline velocity: Number of opps times win rate times average deal size, divided by cycle length. One number for the whole engine. Shortening the cycle compounds; adding raw opps is the slowest lever.
Quota attainment: Bookings against quota, and the share of reps hitting it. Report the share, not the mean: in 2025 the average was ~43% and most reps missed. When most miss, the quota model is guessing, not the reps.
LTV:CAC: Lifetime value over acquisition cost. 3:1 is the rule of thumb: under 3 you burn cash to grow, over 5 you are likely underinvesting. Most people compute it wrong by using revenue instead of gross margin.
Rule of 40: Growth rate plus profit margin should clear 40%. A blunt SaaS health check: it forgives thin margin if you are growing and low growth if you are printing cash.
ACV / ARR / TCV: Annual contract value, annual recurring revenue, total contract value. TCV folds in one-time and multi-year, so a vendor quoting TCV as ARR inflates by the contract length.
Bookings vs revenue vs billings: What you signed, what you recognized, what you invoiced. Three different numbers sales and finance confuse constantly; a booking is not revenue until it is delivered.
Net new ARR / NNARR: New logo plus expansion minus churn and contraction. The one number that grows the business; report it split by motion or it hides a leak.
Logo churn vs revenue churn: Share of customers lost versus share of dollars lost. High logo churn with low dollar churn means you are shedding small accounts, often fine; the reverse is a crisis.
Expansion ARR: Upsell and cross-sell into the existing base. About 2x cheaper per dollar of ARR (~$0.80 versus ~$1.63 to acquire); if NRR is under 100% you are refilling a bucket, not expanding it.
Average deal size / ASP: Total bookings over deals. Moving it up one tier beats adding reps: a 20% ASP lift drops straight through pipeline velocity.
Cost per opportunity / cost per SQL: Marketing and SDR spend over qualified opps produced. The number that says whether pipegen is efficient or you are just buying meetings.
Marketing sourced vs influenced: Pipeline marketing created versus merely touched. Sourced is a clean number; influenced is where attribution fights start, because everything is influenced by something.
Burn multiple: Net burn over net new ARR. Under 1 is elite, 1 to 2 is fine, over 2 means you are spending two dollars to make one of recurring revenue.
Gross margin: Revenue minus cost to deliver. The number that makes SaaS worth more than services; under 70% you have a services business wearing a software valuation.

Data & systems

Source of truth: The one system that owns a given field. The problem is never having a source of truth, it is having three that each think they are it.
Idempotency: A job you can run twice without corrupting data or double-charging. Most ops automations are not idempotent, and that is a data-integrity bug hiding in plain sight.
Order of execution: The sequence automations fire on a record save. Turn one off and the survivors re-order, so a flow that ran third now runs second and reads a field before another has written it.
Reverse ETL: Syncing modeled data from the warehouse back into operational tools. It is plumbing beneath identity resolution and activation, not a dumb pipe; writeback has no undo button.
Field history: The audit trail of who changed what, when. Riddled with gotchas: sandbox history resets, formula fields are not tracked, and there is a hard limit on tracked fields. If inputs are unreliable, dashboards are fiction.
Data decay: B2B contact data goes stale at roughly 22 to 30% a year. Your enriched CRM is quietly rotting; enrichment is a subscription, not a one-time append.
Matching key: The rule that decides two records are the same entity. Email is not unique and company name is not stable; without a durable key (domain, DUNS) you merge two real accounts or split one.
Fuzzy matching: Probabilistic record matching on near-equal strings. Buys you 15 to 20% over exact match and costs you false merges; always set a confidence floor and a human review queue.
Golden record: The merged, surviving version of a deduped entity. It is only golden if survivorship rules are explicit; otherwise the last writer wins, and that is usually the worst source.
Picklist governance: Controlling the allowed values in a dropdown. Free text where a picklist belongs is how you get 43 spellings of United States and a segmentation you cannot trust.
Validation rule: A save-time gate that rejects bad data. Great in moderation; 47 required fields is how reps learn to type junk to get past your form.
Roll-up summary: A parent field that aggregates children, like open pipeline on an account. The gotcha that breaks half of them: it cannot cross a lookup or filter on a formula field.
SCD2 snapshot / slowly changing dimension: History kept by versioning rows with valid-from and valid-to dates. The only honest way to answer what was ARR on Jan 1; a live table cannot.
Data lineage: Tracing a number back through every transform to its source. Without it, why is this dashboard wrong is an all-day investigation instead of a five-minute trace.
Batch vs streaming: Processing on a schedule versus event by event. Most ops does not need streaming; a nightly batch you can re-run beats a real-time pipe you cannot debug.
Sync latency: The lag between a change in one system and its appearance in another. Reps trust the CRM until a 15-minute lag makes them look dumb on a call, then they retreat to spreadsheets.
Integration user: The service account that syncs between systems. Give it least privilege and its own login, or a deactivated employee takes your whole marketing sync down with them.

Outbound & enrichment

Enrichment waterfall: Chaining data providers so a miss from one becomes a hit from the next, and you pay mostly on verified data. A 4-tool waterfall tops out around 68% coverage; anyone quoting 92% is counting catch-alls that bounce.
Catch-all: A domain that accepts any address, so a provider calls it a hit even when the mailbox does not exist. About 9% of domains, 20 to 30% at enterprises. Treat it as risky, not valid.
Signal: An event that marks an account entering a buying window: funding, a key hire, a usage spike. A high-intent list in a spreadsheet is not a signal until it is wired to a workflow.
Deliverability: Whether your mail reaches the inbox at all. A trust system: cross 0.30% spam complaints and providers throttle you. Delivered is not inboxed; 98% delivery can still be 40% placement.
Signal-based outbound: Firing outreach off a real trigger instead of a static list. Signal-triggered replies run 10 to 20% versus 1 to 3% for spray, a 3 to 10x lift from timing alone.
SPF / DKIM / DMARC: The three DNS records that prove you are allowed to send from your domain. Miss DMARC and the Google and Yahoo bulk-sender rules bounce you outright since February 2024.
Domain warmup: Ramping send volume on a new domain so mailbox providers trust it. Skip it and you cook the domain in a week; warmup takes 3 to 4 weeks minimum.
Sending subdomain strategy: Cold outreach from a separate lookalike domain, never your primary. A burn there does not torch your corporate mail; sending cold from the main domain is how a whole company loses the inbox.
Bounce rate: Share of sends that hard-fail. Cross 2 to 3% and providers throttle you; a spike almost always means stale data, not a bad server.
Intent data: Third-party signal that an account is researching your category. Noisy and often stale; treat it as a prioritization tiebreaker, not proof of a buying window.
Technographics: The tools an account already runs, used for fit and displacement plays. Detection is 60 to 80% accurate at best, so no X detected is not does not use X.
Firmographics: Size, industry, geo, and revenue attributes that define fit. The backbone of ICP scoring, and the first thing enrichment gets wrong on private companies.
Cadence / sequence: The scripted multi-step outreach flow. The tool does not make it work; a 12-step cadence to the wrong persona just automates being ignored faster.
List hygiene: Removing invalid, catch-all, and role-based addresses before you send. The cheapest deliverability lever there is; skipping it is why the bounce spike happened.
Suppression list / do-not-contact: Accounts and people you must not email: customers, competitors, opt-outs, GDPR withdrawals. One miss here is a legal problem, not a deliverability one.
Positive reply rate: Interested replies over sends, not total replies. Total reply rate flatters you with out-of-office and unsubscribes; positive over sent is the only number that predicts pipeline.

Lifecycle & handoffs

The bowtie: The recurring-revenue replacement for the funnel: acquisition narrows to Commit, then retention and expansion widen back out. Most orgs instrument the left side heavily and the right side not at all, which is why NRR is a mystery.
MQL / SQL / SAL / SAO: The lead lifecycle stages. Only 8% of companies define them the same way across marketing and sales, which is why the handoff leaks: 53% of MQLs never become SALs.
Handoff: Any point where a record, owner, stage, or responsibility changes. Every handoff is a leakage point: 30 to 50% of B2B revenue leaks at the seams unless you instrument each one.
Speed-to-lead: Time from inbound lead to first outreach. Under 5 minutes is far more likely to qualify than 30. The leak is rarely fairness; it is the median latency nobody measures. 44% of leads are never contacted at all.
Lead-to-account matching / L2A: Tying an inbound lead to the right account before routing. Email-domain matching catches ~70%; fuzzy company-name adds 15 to 20%. Without it, 15 to 25% of leads misroute.
Lead routing: The rules that assign an inbound lead to an owner. Round-robin is fair and dumb; route on territory and account ownership or you split one account across three reps.
Round robin: Distributing leads evenly across a pool. Fair on paper; it ignores capacity and skill, so your best closer gets the same tire-kickers as your newest hire.
Lead SLA: The agreed time-and-touch commitment between marketing and sales on a handed-off lead. Unenforced SLAs are wishes; wire it to an alert and a report or 44% of leads die untouched.
Lifecycle stage: The marketing-side status of a person (subscriber, lead, MQL, customer). Keep it separate from the sales opp stage; conflating person-status and deal-status is a top data-model mistake.
Recycling / nurture return: Sending a not-ready lead back to marketing instead of disqualifying it. Most dead leads are just early; a recycle loop recovers 10 to 20% reps would otherwise trash.
Closed-lost reason: The required disposition on a lost deal. Useful only if the picklist is short and enforced; no-decision hiding 40% of losses tells you nothing.
Opportunity splits: Crediting multiple reps on one deal. Necessary for team selling, a comp nightmare when the rules are vague; define them before the deal, not after.
Segment / ICP tier: Banding accounts by fit and value (enterprise, mid-market, SMB). It drives routing, coverage, and quota, so a fuzzy segment definition breaks all three downstream.

Comp & planning

Pay mix: The base-to-variable split of on-target earnings. It should track how much of the outcome the role controls: AE 50/50 up to 55/45 at enterprise, SDR 65/35, CSM 75/25 to 80/20.
Quota-to-OTE ratio: Quota divided by on-target earnings. 4 to 5x is standard, enterprise runs 5.5 to 6x, and above 6x the plan is structurally broken.
Ramp: The months before a rep carries full quota. If you plan capacity on nominal quota and ignore ramp, you overstate capacity badly: 12 reps at a 6-month ramp carry half of nominal, not all of it.
Capacity planning: Deriving how much a team can actually produce from reps, ramp, quota, and attainment. It is a RevOps problem wearing a finance costume; the inputs all live in the CRM.
Territory design: Carving accounts so every rep gets a fair shot at the number. Most territory fights are really quota fights in a costume; balance the quota-to-pipeline ratio first and half the complaints disappear.
OTE / on-target earnings: Base plus variable at 100% attainment. The number recruiters quote and reps hear as guaranteed; only the base is guaranteed.
Accelerator: A higher commission rate above 100% quota. The lever that keeps A-players from coasting; cap it and your best reps stop selling in December.
Clawback: Reversing paid commission when a deal churns or refunds early. Necessary and hated; a vague clawback window destroys trust, so define it tightly.
SPIF: A short-term cash incentive on a specific product or push. Great for a quarter-end nudge, terrible as a habit; a permanent SPIF is just underpriced base comp.
Draw: Guaranteed pay against future commission, common during ramp. Recoverable draw is a loan reps often do not realize they are taking; non-recoverable is a signing bonus in disguise.
Crediting rules: Who gets credited for a deal, and how much. Crediting disputes eat more RevOps time than rate math does; write the crediting rules before the quarter, not during it.
Comp plan complexity: The number of components a rep must model to know their check. If a rep cannot compute their own commission on a napkin, the plan will not drive behavior.
Quota relief: Reducing a rep quota for a leave, a bad territory, or a comp error. Necessary and abused; without a written policy it becomes a negotiation every quarter.

CS & retention

Health score: A composite of usage, support, sentiment, and payments that predicts renewal risk. Only as good as its inputs; a score built on login count alone shows green on a dying account.
Time to first value / TTFV: How long from signing to the customer getting real value. The best leading indicator of retention there is; a slow TTFV shows up as churn two quarters later.
QBR / quarterly business review: The recurring check-in with a customer. Done right it surfaces expansion and risk; done wrong it is a status meeting the customer starts skipping.
Onboarding: The structured first 30 to 90 days of a new customer. Where most retention is won or lost; a customer who does not reach first value here is already a churn statistic.
Contraction / downsell: A customer staying but shrinking. Quieter than churn and often missed in reporting, and it eats NRR the same way churn does.
Expansion play: A defined motion to grow an existing account: seats, modules, usage tier. Fire it on a usage signal, not a calendar; expanding an account that is not adopting just accelerates churn.
Renewal forecast: The CS analog of a sales forecast, calling which renewals land. Most orgs run it on a spreadsheet and a gut feel; without a health signal it is a hope-based number.
CSM coverage ratio: Accounts or ARR per customer success manager. High-touch runs $2 to 5M per CSM, tech-touch is one-to-many with automation; mismatch it to the segment and you either overspend or lose accounts.
Save play: The scripted intervention when an account goes red. Its whole value is speed; a save play that kicks in at the renewal notice is 60 days too late.
Champion tracking: Watching the internal advocate who sells for you when you are not in the room. When a champion leaves, renewal risk spikes, so track the departure as a signal, not a surprise.
Adoption / product usage: Whether the customer actually uses what they bought. The truest health input; shelfware on their side is your churn at the next renewal.

CPQ & billing

CPQ / configure, price, quote: The system that turns a product selection into a legal quote. Its job is to make reps unable to build an invalid or unprofitable quote; if they still quote in spreadsheets, CPQ failed.
Price book / rate card: The master list of products and their prices. One stale price book is how a rep quotes last year number; version it and gate the edits.
Approval matrix: The tiered sign-off required as a discount deepens. The gate that protects margin; if 40% discounts auto-approve, the matrix is decoration.
Bundle / SKU: A packaged set sold as one line, and the individual sellable unit under it. A clean SKU taxonomy is the difference between reportable revenue and a pricing mess.
Co-term / amendment: Changing a live contract mid-term and aligning the new lines to the original end date. Where proration math and renewal ARR quietly break if the CPQ does it wrong.
Proration: Charging for a partial period when a change lands mid-cycle. The most bug-prone math in billing; a proration error is a customer-trust and a revenue-recognition problem at once.
Ramp deal: A contract with scheduled price or quantity increases over the term. Great for landing a big logo; a forecasting nightmare if your system books the full ramp as day-one ARR.
Usage-based billing / consumption: Charging on what the customer uses, not a flat seat count. Aligns price to value and wrecks predictability; forecasting consumption is a different discipline from seats.
Order form / MSA: The deal-specific pricing document and the master legal agreement under it. The order form sells, the MSA governs; reps who redline the MSA to close faster create the next renewal fight.
Revenue recognition / ASC 606: The accounting rule for when a booking becomes recognized revenue. Why bookings and revenue never match: a three-year prepaid deal is one booking and 36 months of revenue.
List price vs net price: The sticker versus what the customer actually pays after discount. Reporting ARR on list price (a real and common bug) inflates the number by the entire discount stack.

Systems & architecture

Sandbox vs production: The safe copy where you build versus the live org customers touch. Never build in prod; the corollary is that a sandbox with no real data hides the bugs only prod data triggers.
Metadata vs data: The config (fields, flows, layouts) versus the records. Deployments move metadata, so a change that assumes data exists in the target org is how a tested fix breaks on release.
Governor limits: The hard caps a multi-tenant platform enforces (Salesforce CPU, query, DML). You do not opt out; a bulk owner change on a big account hits the CPU limit and fails, sometimes silently.
Declarative vs code / Flow vs Apex: Building with clicks versus writing code. Flows-first is the right default for maintainability; the trap is a 40-element flow doing what 10 lines of Apex should.
Change set / deployment: The mechanism that moves config between orgs. Change sets are slow and manual; the discipline is dry-run first, deploy to a dev org, never straight to prod.
API rate limit: The ceiling on API calls per period. Integrations that do not batch burn through it and take the sync down; the fix is bulk operations, not more retries.
Bulk API: The endpoint built for large data loads in batches. Using the regular API for a 200K-row load is how you hit rate limits; also mind the CRLF line-ending gotcha on some loaders.
Environment strategy: How many orgs you run and what each is for (dev, QA, staging, prod). Too few and you build in prod; too many and nobody knows which is current.
Technical debt: Accumulated shortcuts that make the next change harder. In GTM systems it shows up as 282 automations nobody dares turn off, and interest is paid on every future deploy.
Custom vs standard object: Extending the platform with your own object versus using the built-in one. Reach for standard first; a custom object you could have modeled as a field is debt you chose.
Middleware / iPaaS: The integration layer (Workato, Mulesoft, Zapier) connecting systems. Good middleware is documented and monitored; bad middleware is a black box that becomes an undocumented source of truth.

Security & compliance

SOC 2: The audit proving you handle customer data responsibly (Type I is a point in time, Type II covers 3 to 12 months). Enterprise buyers gate the deal on it: no report, no security review, no signature.
SBOM / software bill of materials: The ingredient list of every component in your software. Post-Log4j and post-XZ it is table stakes; you cannot patch a vulnerability in a dependency you cannot see.
Least privilege: Giving every user and service account the minimum access needed. Most CRMs default to the opposite, which is why one phished admin can own the entire org.
SSO / SAML: Single sign-on so identity lives in one place (Okta, Entra) instead of per-app passwords. The SSO tax is real: many vendors gate it behind the top tier to force enterprise deals.
SCIM: The standard that auto-provisions and deprovisions users from your identity provider. Without it, deactivated employees keep CRM access for months; SCIM closes that gap the moment HR does.
Data residency: The requirement that data physically stays in a region (EU, US). A hard gate for EMEA and public-sector deals; we can spin up an EU instance is a roadmap item, not a yes.
PII / PHI: Personally identifiable and protected health information. The categories that trigger GDPR, CCPA, and HIPAA; dropping them in a free-text CRM note is a breach waiting to happen.
Audit log: The immutable record of who did what in a system. The first thing an auditor asks for, and the first thing a badly governed org cannot produce.
Field-level security / FLS: Permissions on individual fields, not just objects. The correct way to hide comp or SSN data; do it in profiles and permission sets, not by hiding a column in one report.
RBAC / role-based access control: Granting permissions by role rather than per person. It scales; the per-user alternative becomes an un-auditable mess by headcount 50.
Vendor security review / security questionnaire: The questionnaire and data-processing agreement a buyer security team runs before signing. Budget 2 to 6 weeks; the deal is not closed until security clears it.

AI & data

RAG / retrieval-augmented generation: Grounding an LLM in your own documents instead of its training data. The default pattern for GTM AI; without it the model confidently invents your pricing.
Hallucination: An LLM stating something false with full confidence. The reason raw model output never touches a customer without a human or a grounding source; a hallucinated price is a real liability.
Prompt injection: Hostile text hidden in data that hijacks an LLM instructions. The new SQL injection: if your agent reads inbound email, someone will try to reprogram it through one.
Embedding / vector search: Turning text into numbers so you can find semantically similar records. The engine under RAG and fuzzy account matching; similar is not correct, so set a similarity threshold.
Fine-tuning vs prompting vs RAG: Three ways to make a model do your task. Most GTM problems are a prompting or RAG problem; fine-tuning is the expensive answer people reach for too early.
Agent: An LLM that takes actions, not just answers: updates a record, sends an email. Powerful and dangerous; an agent with write access and no rollback is how the agent graveyard fills up.
Human in the loop: A required human approval before an AI action commits. The single control that keeps pilots alive past 90 days; remove it before trust is earned and one bad batch ends the program.
Context window / tokens: The amount of text an LLM can hold at once, and the unit it is billed on. Cost and truncation both scale with it; stuff too much in and the model quietly forgets the middle.
Model eval: The labeled test set that catches quality drift over time. No evals means you find out from a customer; a small ground-truth set is the cheapest insurance in AI ops.
Ground truth / labeled data: The verified-correct answers you measure a model against. AI ops lives or dies on it; without ground truth, accuracy is a vibe.
Structured output / function calling: Forcing an LLM to return clean JSON or call a defined tool. The line between a demo and a production pipeline; free-text output breaks every downstream integration.

The slang

Sandbagging: A rep hiding upside to beat a soft number later. When commit accuracy is above 90%, that is not discipline, it is sandbagging.
Zombie deal: An opportunity that will never close but still sits in the pipeline inflating coverage. A 4x pipeline with 30% zombies is really 2.8x.
Agent graveyard: The pile of AI pilots that died at 90 days because no one owned them, the data was dirty, and there was no rollback. 88% of teams use AI; only 38% scale past the pilot.
Automation debt: The accrued weight of redundant, dormant, and self-fighting automations. It charges interest in save latency and order-of-execution bugs. One real org ran faster after cutting 282 automations to 82.
Dumpster fire: The state of a CRM instance nobody governed for three years: duplicate rules set to block, 47 required fields, six flows writing the same owner. The affectionate term for the thing RevOps is hired to put out.
Hockey stick: The forecast shape where nothing closes until the last two weeks of the quarter, then everything supposedly does. Usually a slippage problem wearing optimism.
Happy ears: A rep hearing a buying signal that is not there. The forecast tax on optimism, and the reason deep discovery beats deal excitement every time.
Commit theater: The ritual of calling a commit number everyone in the room already knows is soft. When the real number is common knowledge, the forecast meeting is a performance.
Spray and pray: Blasting a huge list with no targeting and hoping for replies. 1 to 3% reply if you are lucky, and you torch domain reputation doing it.
Franken-stack: Fifteen tools bolted together with duct tape and Zapier. Every one promised to consolidate the stack; each added a sync job and a new source of truth.
Swivel-chair integration: A sync that is really a human copying data between two browser tabs. The most common integration in GTM and the least documented; it breaks the day that person takes PTO.
Garbage in, gospel out: Bad data laundered into a confident dashboard nobody questions. The dashboard does not make the number true, it makes the number unquestioned.
Shelfware: Licenses paid for and never used. The average company wastes 30%+ of SaaS spend on it; the CRM add-on nobody adopted is the classic.
Data janitor: The RevOps person whose real job is cleaning up after everyone else inputs. Half the role, and the half nobody put in the job description.
Forecast whiplash: The called number swinging 30% week to week because the process is guessing. That is not a forecast, it is a mood ring.
Tool sprawl: The unchecked growth of overlapping point solutions. Every team buys its own; RevOps inherits the reconciliation bill.