Data enrichment mistakes: 12 anti-patterns I've seen kill projects

The fastest way to burn enrichment credits is to re-enrich records that already have the field you're looking for. It sounds obvious — until you realize that every "refresh all" script does exactly this by default. A naive nightly job that calls the enrichment API on every CRM contact will spend 70-80% of its credits on records that haven't changed.

The fix: diff before you enrich. Only call the API on records where the target field is null, or where it was last updated more than X days ago. A simple WHERE clause cuts credit consumption by 5-10x with zero quality loss. Most teams discover this after their second month's bill.

How Derrick handles it: our Sheet integration auto-skips rows that already have the requested field populated (unless you explicitly toggle "force refresh"). Same with the API/MCP — idempotent by default. You only pay for new info.

Enrichment APIs return emails with a "confidence" or "validity" score. The score is calibrated against the API's own ground truth — not against what mailgun or SES will accept. A provider can claim 90% "valid" emails while 20-25% of those bounce in production.

The damage compounds: bounces hurt sender reputation, which hurts inbox placement, which hurts reply rate on the emails that DID land. A typical outbound team that skips verification loses 30-40% of expected reply volume to deliverability issues they don't attribute correctly — they blame the copy instead.

The fix: always verify emails before sending. The cost is ~€0.005/email (one-tenth the cost of enrichment itself) and saves your domain reputation. Pair it with bounce monitoring so you can pull a provider that suddenly drops in quality.

How Derrick handles it: verification happens in the same call as enrichment, with a separate "validity" column you can filter on before exporting to your outreach tool. One step, not two, and no "forgot to verify" mishaps.

Teams routinely run enrichment on a dirty CRM: duplicate contacts, mismatched company-account links, normalized job titles that didn't get normalized. The enrichment then adds new fields on top of records that should have been merged first. You end up with the same person enriched 3 times under 3 contact IDs — paying 3x and creating worse data, not better.

The discipline: cleanse before you enrich. Dedupe by email, normalize company names to a single account, fix obviously stale records ("CEO of company that died in 2019"). Then enrich. The order matters: enrichment is multiplicative on quality, so if you enrich dirty data, you get cleaner-looking dirty data — which is worse than obviously dirty data because nobody knows to question it.

How Derrick handles it: we surface duplicate-email and company-mismatch warnings before enriching, so you can dedupe in the same Sheet pass. No separate cleansing tool required for the basics.

Enrichment everything in the CRM "just in case" is one of the most expensive patterns in B2B. A typical RevOps team that enriches all 50k leads in the CRM monthly is spending €5-8k/month, when only ~5k of those leads will ever get touched by a rep. That's a 10x waste ratio sitting in plain sight.

The fix: enrich on the trigger that creates value — when a rep is about to call, when a sequence is about to start, when a deal moves stage. Not on a schedule that ignores whether anyone needs the data. Pair this with cached results: if Derrick enriched a contact 30 days ago and the rep finally opens it today, return the cache, don't re-charge.

How Derrick handles it: trigger-based enrichment is the default mode in the Sheet — enrich a row when you mark it for outreach, not on every refresh. And the cache returns previous enrichments for free, so reps can re-open without re-billing.

An enrichment program left on autopilot drifts. Match rates slowly degrade as providers change their data sourcing. Cost per match creeps up as ICPs evolve and the easy matches get exhausted. Compliance posture goes stale as regulations tighten. By month 12, the program your CFO approved at quarter 1 is meaningfully worse — but nobody notices because the system "works."

The fix is operational, not technical: quarterly audit, 30 minutes, four checks. (1) Match rate per provider trend. (2) Cost per verified match trend. (3) Bounce rate trend. (4) Provider compliance status (DPA still valid, GDPR basis still documented). Four numbers, four trend lines. If any goes the wrong way for two consecutive quarters, you have a fixable problem before it becomes a CFO-meeting-problem.

How Derrick handles it: all four metrics live in the Sheet dashboard, refreshed automatically. The quarterly audit is reading 4 numbers off your existing setup — 5 minutes, not 30.