Account scoring to a BDR action layer

Built from the system Garrett Wolfe, founder at 1GTM, described on GTM Vault SMYS 3 (https://pro.gtmvault.co/episodes/smys-03). Published free by GTM Vault. Numbers below are the operator's own, from the episode; your results will differ.

When to use this

Use this when a company has a CRM, a defined TAM and a BDR or sales team, and the reps decide which accounts to work by preference and then spend their day moving between tools to find people and contact data. It fits teams that already see intent signals but stop at the company level and leave the rep to work out who to call. Garrett builds it for B2B SaaS clients and for services businesses in the legal and PE space (SMYS 3, 2:04). The second half of the system assumes someone can build a simple web app on the team's APIs, which he did with Claude Code (SMYS 3, 3:07).

The system on the record

Garrett's view is that sales and BDR teams need bespoke tooling that fits their workflow, and that the highest leverage is making those people more efficient, on a foundation of data first (SMYS 3, 2:04). The flow ingests every account in the TAM, scores and prioritizes it on criteria the client cares about, such as headcount, revenue, certain types of leadership and recent news announcements (SMYS 3, 1:02), then finds people at those accounts, finds reasons to reach out, and sends email and LinkedIn outreach (SMYS 3, 2:04). It runs in five parts:

1. Account scoring. He pulls tens of thousands of companies into a Clay table and merges the firmographic and technographic fields the client cares about (SMYS 3, 3:07). Each account is classified by industry and the industry is scored. A data waterfall starting with Lead Magic pulls revenue, LinkedIn and employee count. Every criterion is scored on a continuum of 0 to 1, the account gets an overall score, and scores are sorted into tiers. The signals he scores in the demo include hiring, revenue, engineering count and infrastructure (SMYS 3, 3:53). The point is to decide which accounts matter and why, rather than have BDRs pick their 10 favorite accounts for the week (SMYS 3, 4:57).
2. Signal aggregation on scored accounts only. He runs signals on a subset of the TAM, for example the top 60% to 70% of accounts, or companies that scored better than 40% overall (SMYS 3, 4:57). In the demo he uses Clay's native new hire signal, filtered to the accounts selected by score, with each row a person carrying the account score (SMYS 3, 5:47).
3. Qualification and contact data. An OpenAI model qualifies or disqualifies each person, qualified people are mapped to a persona, and a data waterfall runs across several vendors with an email find node and a validation node to produce validated work emails (SMYS 3, 5:47). Names are normalized and a personalized first sentence is written (SMYS 3, 6:49).
4. Routing. People go into two campaigns based on persona, one of them in HeyReach for LinkedIn automation. The team is notified when the score is above a set threshold, and in some cases the data goes to Google Sheets (SMYS 3, 6:49). Before warm outbound, the workflow checks whether the person is at an account that should be disqualified because it is a current customer or a prospect, confirms they are the right person, gets their contact information, drops them into outbound workflows and syncs back to the CRM (SMYS 3, 6:49).
5. Intent HQ, the action layer. Because automated outbound tops out, he builds a custom web app on top of the pipeline that gives BDRs one action-oriented place, instead of jumping between the CRM, Sales Navigator and Apollo to work out which accounts they own and who to call (SMYS 3, 7:51). The aim is to use the existing infrastructure and abstract away the work of using LinkedIn and Apollo (SMYS 3, 8:53). The app has these parts:
   1. Company view. It pulls the client's CRM data, not net new enrichment, and shows existing signals with dates, such as a fundraise or someone from the company following the client's page. Reps filter by signal, company type and country to build a high fit list with a talk track tied to the signal (SMYS 3, 9:56).
   2. People tab. It shows person-level intent, such as named website visitors, with a verified email and an option to enrich for a phone number (SMYS 3, 11:00). A rep can curate their own list, or a RevOps or BizOps person can cut lists and deliver them (SMYS 3, 11:45).
   3. Prospecting sheet. Reps select people, enrich missing details, and export or sync them. Data syncs both ways with the CRM, and the rep can download a CSV for whatever dialer the team uses, with the fields already filled (SMYS 3, 12:41).
   4. One-click prospecting. Where a signal exists only at the company level, a prospect button finds people at that account and enriches them, so the rep does not run Sales Navigator and Apollo separately (SMYS 3, 13:45 and 14:36). In the demo the button failed on two accounts (SMYS 3, 13:45). It shows only people who match pre-baked personas, not everyone at the company; for his client Antimetal at an account like Meta, that means people in site reliability, infrastructure or platform roles (SMYS 3, 17:15 and 18:09). People found this way are net new to the CRM and go onto a calling list that syncs back (SMYS 3, 16:21).

He is explicit that Intent HQ does not replace the CRM. The CRM stays the source of truth, and the app is a simple action layer for business and sales development reps (SMYS 3, 17:15). He built it with Claude Code and a working knowledge of the APIs in the stack (SMYS 3, 3:07 and 22:07).

The result

1. Reply rates on the automated workflows: he puts the practice at 1% to 12% reply rates at best, including out of office replies (SMYS 3, 7:51).
2. Meetings: the automated outbound workflows generate "dozens" of meetings per week and per month across all of his clients. He gave no exact count (SMYS 3, 19:12).
3. Time saved per rep with Intent HQ: when asked for results, he said "as little as like half an hour to an hour saved per day" across his clients (SMYS 3, 19:12 and 20:14). Earlier in the episode he estimated it saves "at least like two hours a day" for many BDRs in heavy prospecting and sourcing roles (SMYS 3, 11:45). The two figures differ and he did not reconcile them.
4. First-day example: on the first day one client used the tool, the sales team booked a meeting with a Fortune 500 company within about five minutes on a parallel dialer (SMYS 3, 20:14).
5. He gave no before and after figures for conversion lift, pipeline generated or cost reduced from Intent HQ. His email funnel figures (100 sent, 20 delivered, 10 opened, 8 clicked, 5 replied) were an illustration of email as a numbers game, not a measured result (SMYS 3, 19:12).

Run it on your company

1. Ask once, in one message, for: the CRM in use and API access to it; the TAM source and its approximate account count; the scoring criteria the team cares about (headcount, revenue, leadership types, recent news, hiring, technographics); the enrichment vendors available for a waterfall; the signals already captured (new hires, fundraises, job posts, page follows, website visitors); the personas the team sells to; the outbound channels and sequencer; the dialer; how many BDRs there are and how they pick accounts today; and who will build and maintain the web app.
2. Write the scoring model. Map each criterion to a field, a source and a 0 to 1 scoring rule, then combine into an overall score and tiers. Output: a table with columns Criterion, Field, Source (with waterfall order), Scoring rule (0 to 1), Weight, plus a tier table with columns Tier, Score range, Treatment. Leave weights for the user to confirm; Garrett gave none on the record.
3. Set the signal gate. Pick the cutoff that decides which accounts get signal monitoring, using Garrett's examples (top 60% to 70% of the TAM, or an overall score better than 40%) as a starting point. Output: the cutoff, the resulting account count, and a table with columns Signal, Source, Level (company or person), Refresh frequency.
4. Write the qualification and contact flow. Output: a fenced qualification prompt that takes one person row and returns qualified or disqualified with a reason and a persona; a persona table with columns Persona, Titles, Campaign; a contact waterfall with columns Order, Vendor, Step (find or validate); and the disqualification checks for current customers and open prospects.
5. Design the routing. Output: a table with columns Persona, Channel, Campaign, Notification threshold, CRM sync field, plus the personalized first sentence prompt as a fenced block with three sample outputs for user review.
6. Spec the action layer. Define the views Garrett showed: company view with dated signals and filters, people tab with person-level intent, prospecting sheet with enrich, export, CRM sync and dialer CSV, and one-click prospecting limited to persona matches. Output: a screen-by-screen spec with columns Screen, Data shown, Source (CRM or API), Actions, Writes back to CRM (yes or no), plus a list of the APIs each action calls. State that the CRM remains the source of truth and the app reads from and writes to it.
7. Plan the build. Output: a step list for building the app with Claude Code or the team's own tooling, starting from a sandbox on example CRM data, with a test plan for each action that calls an outside API.
8. Set the scorecard: reply rates and meetings from automated outbound, time spent prospecting per rep per day before and after the app, meetings booked from Intent HQ lists, and net new people added to the CRM through one-click prospecting, reviewed weekly.

Tools on the record

1. Clay: account scoring and enrichment waterfall.
2. Lead Magic: first call in the enrichment waterfall.
3. Apollo: contact database.
4. LinkedIn Sales Navigator: research and prospecting.
5. HeyReach: LinkedIn automation path in two-path campaign routing.
6. Claude Code: built the Intent HQ action layer.

Any tool doing the same job works; the method is the point. Full stack by layer: https://pro.gtmvault.co/stack/operator/garrett-wolfe

Where it breaks

1. No scoring foundation. Without a scored TAM, reps go back to picking their 10 favorite accounts for the week, and signals arrive on accounts that were never a fit (SMYS 3, 4:57).
2. Intent that stops at the company. Garrett names platforms that report an intent signal for a company and leave the rep to find people as the gap Intent HQ closes (SMYS 3, 13:45). A company-level signal still needs a person to call or email (SMYS 3, 17:15).
3. Another tool reps ignore. He frames the goal as giving reps data that is not just a CSV or a dashboard they are not going to use (SMYS 3, 4:57). An app that does not fit the daily workflow repeats that problem.
4. Replacing the CRM. He keeps the CRM as the source of truth and the app as an action layer on top (SMYS 3, 17:15). An app that holds its own records breaks the two-way sync the prospecting sheet depends on (SMYS 3, 12:41).
5. Custom code needs upkeep. The one-click prospect button failed on two accounts during the live demo (SMYS 3, 13:45). A bespoke app built on several APIs needs an owner and testing.
6. Expecting automation alone to carry pipeline. He puts automated outbound at 1% to 12% reply rates at best, including out of office replies, which is why he builds the action layer for the sales team on top (SMYS 3, 7:51).