Straight answers, real numbers

Scrape when the data you need isn't in any database you can buy: a niche or data-poor market, a field no vendor sells, or freshness a static file can't match. For mainstream, well-covered B2B, buying is faster and cheaper; custom scraping is the right tool in maybe 10–20% of cases. And it's a commitment, not a one-off: a scraper costs roughly 20–30% of its build in maintenance every year as sites and anti-bot defenses change. Scrape for the gap, buy for the bulk — and never scrape what's already a column you can purchase.

GTM engineering is the practice of building the data, software, and automation systems behind a go-to-market motion — instead of throwing more headcount at it. A GTM engineer maps the TAM, sources and scores the data, wires the CRM, and automates the busywork that consumes reps. It's RevOps with a builder's toolkit: reverse-engineer what's already closing and turn it into a repeatable system. The evidence favours building for people rather than around them — teams using AI to augment reps see roughly 2.8× the pipeline of teams trying to replace them, and 81% of sales teams were investing in AI by 2025.

Yes, when it's fit on your own closed-won outcomes rather than on guessed point values — and when the data underneath it is complete enough to rank. That second condition is where most scoring quietly fails: B2B records decay about 30% a year and single-source enrichment plateaus near 78–84% match, so a model reading half-empty, stale records is ranking noise with impressive-looking confidence. Build it on real outcomes, validate it on held-out history by reporting lift by decile, and write the reasons back alongside the score. Skip any of those and you get a number reps learn to ignore.

Waterfall enrichment chains data providers in sequence: the first source fills what it can, the next takes the misses, then the next, until coverage is maxed. It exists because single-source data has a hard ceiling — contact accuracy plateaus near 78–84%, and one database has no fallback when a record isn't in its index. A well-configured waterfall regularly pushes match rates past 90% and lifts direct-dial coverage 20–40% above any single provider. The gain comes from sequencing, not from the number of providers.

B2B contact data decays about 30% a year. The biggest driver is job change: 65.8% of contacts change title or function within 12 months, and 42.9% change phone number in a year. That's why a list is only as good as the day you pulled it, and why a CRM left to drift becomes actively misleading rather than merely incomplete — a wrong field is worse than an empty one, because nobody checks it. Fighting decay means re-running enrichment and verification on a schedule, not buying a static file once.

B2B data enrichment costs roughly $0.01 to $2.50 per record, with high-volume tools near a penny and enterprise providers above $1.50; annual contracts run $12k–$80k. But sticker price is misleading. The real figure is effective cost: list price divided by match rate divided by accuracy — because a cheap record that doesn't match, or matches wrong, costs more than a verified one once it wastes rep time and pollutes your CRM. Calculate it per segment, not in aggregate, or the well-covered majority will hide the expensive tail.

Judge a sourcing partner on four things: whether they'll tell you when you don't need them, how they handle maintenance (commonly 20–30% of build cost per year), whether every record carries provenance, and whether they quote effective cost rather than per-record price. Custom sourcing is the right tool for maybe 10–20% of a data need — the genuine gaps — so a partner who scopes your whole market as custom capture is selling you something you can buy cheaper. Managed extraction commonly clears 90%+ success where ad-hoc DIY efforts sit near half that; ask which one you're buying.

Done-for-you enrichment means someone else owns the whole chain: provider selection and sequencing, verification thresholds, conflict resolution, delivery into your CRM, and the refresh cadence that fights roughly 30% annual decay. Buying a tool gets you the plumbing; the result still depends on configuration nobody does for you. The gap is measurable — single-source enrichment plateaus near 78–84% match while a well-built chain clears 90% and lifts direct-dial coverage 20–40%. Same providers, different outcome, because the sequencing and verification are the product.