If you're still setting up your first waterfall, start with our step-by-step waterfall guide and the complete Clay playbook — this post assumes you know your way around a table. The problem with most "Clay use cases" articles is that they're menus, not recipes: workflow names with no columns, no prompts, no costs. Here are ten builds we actually run, each with the setup, the per-row economics, and the failure modes. Prices and features verified July 2026 (post-March-2026 pricing: Launch from ~$167/mo annual, Growth from ~$446/mo — full breakdown in our Clay pricing review).
| # | Workflow | Approx. cost per row | Plan floor |
|---|---|---|---|
| 1 | Navigator gated-directory scraping | 3 credits (Argon/Navigator) | Launch |
| 2 | Competitor G2 reviewer poaching | ~5–7 credits (research + waterfall) | Launch |
| 3 | Free jobs-data hiring-signal engine | 0 credits for the signal itself | Free |
| 4 | Website visitor → warm outbound loop | ~2–4 credits | Launch (+RB2B/Vector) |
| 5 | Job-change champion tracking | ~3–5 credits per trigger | Launch |
| 6 | Self-healing CRM enrichment loop | ~1–3 credits | Growth (CRM auto-sync) |
| 7 | BYO-API-key credit arbitrage | Direct API price (below credit rates) | Growth (HTTP API) |
| 8 | Multi-table qualification pipeline | 2 credits (Neon) + downstream only on qualified rows | Launch |
| 9 | Signal-stacked scoring (AI formulas) | 0 credits for scoring (formulas are free) | Launch |
| 10 | Deep personalization column stack | ~4–5 credits + ~6 actions | Launch |
Navigator (launched late 2025) is a browser-using agent — it fills search forms, clicks filters, and extracts from JavaScript-heavy or gated directories that defeat normal scraping. Setup: input URL column → Navigator column with explicit interaction steps ("search for {keyword}, apply the {industry} filter, extract the results table as JSON") → split the JSON into columns. At 3 credits per row, prototype on 10 rows in Sandbox Mode first. Best targets: association member directories, conference attendee lists, competitor pricing pages.
Claygent scrapes a competitor's G2 reviews → filter to 1–2 star ratings → a second Claygent finds the reviewer's LinkedIn and company → email waterfall → AI column writes an opener referencing the specific pain in the review. This is the highest-relevance cold angle that exists: the prospect has publicly documented the problem you solve. Keep the send cohorts small and the copy restrained — reference the frustration category, never quote the review verbatim.
Clay indexes roughly 18 million LinkedIn job posts, queryable at zero data-credit cost — the most underrated free signal on the platform. Source: job postings matching pain keywords (e.g., "hiring SDRs," "RevOps manager") → dedupe by company domain → enrich decision-makers → free AI formula qualifies ("does this JD imply in-house outbound? Y/N + evidence"). A company hiring three SDRs is telling you its pipeline strategy; this pairs directly with our signal-based prospecting playbook.
Clay added native RB2B and Vector integrations in April 2026. Pipe visitor identification into a Clay table via webhook → enrich and verify → formula-score on fit (size, title match) → route: high scores go to a Smartlead/Instantly sequence plus a Slack alert; group rows by domain so two visitors from one account escalate priority. Guardrails: enable hot-leads-only at the source and add dedupe rules before the webhook, or repeat visitors will flood the table. More on intent routing in our intent data guide.
Import closed-won contacts and past champions → Signals hub → "Monitor job changes" → on trigger: waterfall the new work email, Claygent researches the new company's fit, Slack alert with an auto-drafted note referencing the prior relationship, write-back to CRM. Champions who move are your warmest possible outbound — reply rates on these campaigns routinely run several times cold baseline (cold average is just 3.43% in 2026 per Instantly's benchmark; see our benchmarks breakdown).
HubSpot Lookup as a source on an auto-update schedule → carry the HubSpot Record ID on every row → enrich, score, dedupe → Update action keyed on record ID with "Ignore blank values" ON — the single setting that prevents Clay from overwriting populated CRM fields with nulls. Run nightly. This turns data decay (B2B data rots ~2–3% a month) into a solved background process.
HTTP API columns calling Prospeo, FullEnrich, or LeadMagic directly with your own API keys, chained with conditional run settings ("only run if previous column is empty") — you rebuild the waterfall at direct-API prices instead of Clay's marketplace rates. Sculptor now auto-configures the endpoint, auth headers, and request body from a plain-English description, which removes most of the technical barrier this tactic used to have. Run the math for your volume: at tens of thousands of enrichments a month, the savings pay for the Growth tier by themselves.
Never run expensive enrichment on unqualified rows. Table 1: TAM companies (CPJ search or Lookalikes seeded from your customer list) → Claygent Neon qualification (2 credits) → send only qualified rows to Table 2 for contact finding and waterfalls → Table 3 for copy → sequencer push. Clay's Functions feature (March 2026) lets you define the scoring logic once and reuse it across every table. Qualify-then-enrich typically cuts total credit burn 40–60% versus enrich-everything.
AI formula columns cost zero credits — use them for all conditional logic. Stack three or more signals per account (funding, hiring velocity from workflow #3, technographics, web intent, LinkedIn activity) → weighted score via formula → tier into cohorts of ≤50 for segmented sends (small cohorts reply 2.76× better than 1,000+ blasts) → route tiers to different campaigns, with email non-responders flowing to a HeyReach LinkedIn campaign. The scoring model itself is covered in our lead-scoring guide.
One research column per angle (recent LinkedIn posts, news, job descriptions, case-study match) → one AI snippet column per angle with a SKIP fallback prompt pattern: "Write one sentence referencing {signal}. If the signal is vague or missing, return exactly: SKIP." → a free formula column picks the best non-SKIP snippet → push as custom variables to your sequencer. The SKIP pattern is what separates usable AI personalization from the generic slop prospects now delete on sight — depth of personalization lifts replies 32–142% depending on how specific the signal is. Total cost: ~4–5 credits and ~6 actions per row, roughly $40–60 per 1,000 fully personalized leads at Growth-tier rates.
In March 2026 Clay moved to a dual system: Data Credits (marketplace data purchases) plus Actions (platform operations like AI columns, HTTP requests, and exports). Launch runs ~$167/mo annual with 3,000 data credits + 15,000 actions; Growth ~$446/mo with 6,000 + 40,000. Unused data credits roll over up to 2× your monthly amount. Full analysis in our Clay pricing review.
A browser-using AI agent inside Clay that interacts with websites the way a human does — filling forms, clicking filters, paginating — so it can extract data from JavaScript-heavy or gated sites that block conventional scrapers. It costs 3 credits per row, so reserve it for sources you genuinely can't get any other way.
Most of the ten run on Launch. Two are gated to Growth: anything using HTTP API columns (workflow 7) and native CRM auto-sync (workflow 6). The free plan's 200-row table limit makes it a testing ground only.
Yes — Clay pushes to Smartlead, Instantly, HeyReach, Woodpecker, and most major sequencers natively, mapping enriched columns to custom variables. Tool trade-offs are in our cold email software comparison.
Only with quality control. Average reply rates fell to 3.43% in 2026 largely because of undifferentiated AI outreach — but specific, signal-grounded personalization still roughly doubles replies versus generic copy. The SKIP-fallback pattern in workflow 10 is the difference: send nothing rather than send filler.
Want these workflows built and run for you? GenFlows designs Clay systems like these — signal engines, enrichment loops, and personalization stacks — as part of complete outbound programs. See the full Clay playbook or talk to our team.
By the GenFlows GTM engineering team. Features and pricing verified July 2026. Last updated July 2026.