Account monitoring universe

Built from the system Umar Farooq Adam, who runs GTM programs globally at Hitachi Vantara, described on GTM Vault SMYS 5 (https://pro.gtmvault.co/episodes/smys-05). 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 the target account list is far larger than the team can work and reps still pick accounts by instinct each week. It fits enterprise and project-based sellers whose buying windows open on a structural event rather than a renewal date. The output replaces "everyone on the list gets sequenced" with "everyone on the list gets watched, and only signal-bearing accounts get touched".

The system on the record

Umar built this for a global systems integrator client pushing into EMEA on sovereign cloud and cloud repatriation. The client has a TAM of more than 150,000 enterprise accounts and reaches out to the top 300 every month (SMYS 5, 1:53). Setup took the first two months of a roughly six-month engagement (SMYS 5, 1:53).

First-party data from billing, support and partners, Bombora intent, CRM activity, Outreach engagement and marketing automation data are all ingested into Clay (SMYS 5, 2:53). Clay then produces four independent scores:

1. Firmographic. ZoomInfo, HG Insights and native Clay enrichment run as a waterfall for employee count and revenue. A small Clay agent classifies industry as focus, adjacent or other (SMYS 5, 5:32).
2. Arbitrary. Estimated IT budget, taken as a share of revenue by industry, and an IT storage estimate from a Clay agent. He calls these arbitrary because they are estimates (SMYS 5, 5:32, 6:34).
3. Technographic. The heavy lift. One of only three mainframe suppliers is exiting that business, so any account running a mainframe attached to a cloud, or an open system attached to a mainframe, is the low-hanging target. HG Insights shows when a product was first identified, its lifecycle and whether it is due for refresh. Mid-range storage and current cloud vendors are scored too (SMYS 5, 3:40, 6:34).
4. Sales play readiness. Another Clay agent researches quarterly reports and public appearances to pick which sales play suits the account best (SMYS 5, 6:34, 7:37).

The four scores roll up into an account score from 0 to 100, returned with an account brief naming the best-suited play (SMYS 5, 7:37).

The scored list is not the trigger. In his words: "We are not reaching out to everyone in that account list." (SMYS 5, 4:27). The system watches four signals across the list: website visits, recent job changes, mentions in quarterly reports and consistent news coverage (SMYS 5, 4:27). From there each stage gates the next:

1. An account trips a signal.
2. Clay writes a personalized message and sends it through the Outreach API, on LinkedIn and email (SMYS 5, 4:27, 8:39).
3. Activity (an open, a reply, or the email seen more than usual) is fetched from the Outreach API and pushed back into Clay (SMYS 5, 4:27).
4. A webhook notifies the right rep on Slack or Teams, and only engaged accounts get a personalized phone call (SMYS 5, 8:39, 10:44).

Inbound runs off the same list: ad audiences and vendor event targeting use the scored accounts, and both motions route through Outreach into Chili Piper for meetings (SMYS 5, 4:27). Because the client hands accounts to preferred local partners who make calls and visits, getting the targeting right is the whole job (SMYS 5, 9:40).

The result

1. Inbound lead time went from 17 to 18 days to the same day (SMYS 5, 10:44).
2. Cost went from a London research firm at 300,000 pounds a year to less than 20,000 a quarter (SMYS 5, 9:40).
3. The client had been heading toward hiring 40 additional sales reps and dropped that path for automation (SMYS 5, 8:39, 9:40).

Run it on your company

1. Ask the user once, in one message, for: the ICP and focus versus adjacent industries; TAM size and where the account list lives; how many accounts the team can genuinely work per month; enrichment and intent sources they own; the structural event that opens buying windows in their market (their equivalent of the mainframe exit); their defined sales plays; the CRM, sequencing tool, meeting router and chat tool (Slack or Teams); and the current average time from meeting request to booked meeting.
2. Name the buying window. Write one paragraph stating the event, the installed base it affects and the technographic fields that reveal exposure. If the user cannot name one, stop and say the technographic layer will be weak and the score will collapse into fit.
3. Produce the scoring spec as a markdown table with columns: Layer, Field, Source order (waterfall), Rule, Points, Layer weight. Four layers: Firmographic (employee count, revenue, industry class focus/adjacent/other), Derived estimates (budget as a share of revenue using a benchmark the user supplies, plus one capacity estimate relevant to the product), Technographic (window exposure, lifecycle or refresh status, adjacent products, current vendors), Sales play readiness (best-fit play and match strength). Weights sum to 100. Mark derived estimates as estimates, used for magnitude, not precision.
4. Write the sales play research agent prompt. Inputs: domain, latest quarterly or annual report, recent public appearances and news. Output JSON: {"best_play", "evidence" (two cited sentences), "match_score" 0 to 100, "brief" (three lines a rep reads before calling)}. Constrain best_play to the user's play list only.
5. Set the monitoring universe cutoff. Size it so that expected signal volume across the universe roughly matches monthly working capacity from step 1. Write the cutoff score and the resulting account count.
6. Define the watch list as a table: Signal, Source, Freshness window, Action. Start from the operator's four (website visit, job change, quarterly report mention, sustained news coverage), add first-party signals the user has, and remove any they cannot capture.
7. Write the gate table: Stage, Entry condition, Channel, Owner. Stages: In universe, Signal fired, Personalized message (LinkedIn and email), Engaged (open, reply, repeat views), Phone call, Meeting booked. No account reaches a call without passing every prior gate.
8. Spec the activity loop: which sequencing API endpoint is polled, what field lands back on the account row, what webhook fires, and the exact rep alert text. Template: "[Account] [score] | [signal] [time ago] | Play: [best_play] | Brief: [brief] | Call now."
9. Point inbound at the same list: export the universe as the ad audience and the event target list, refreshed on the same cadence as scores.
10. Deliver one markdown document, "Monitoring Universe Spec", containing steps 2 to 9, plus a measurement line: baseline request-to-meeting time today, and the date to re-measure.

Tools on the record

1. ZoomInfo: first source in the firmographic waterfall.
2. HG Insights: firmographic waterfall alongside ZoomInfo and Clay.
3. Clay: account research and enrichment.
4. Bombora: third-party intent ingested into Clay.
5. Salesforce: enterprise system of record.
6. HubSpot: contact and account data.
7. Marketo: marketing automation fed from the scored account list.
8. Adobe Journey Optimizer: marketing automation alongside Marketo.
9. Outreach: outbound sequencing.
10. Slack: rep notification fired by Outreach activity webhooks.
11. Chili Piper: meeting routing and scheduling.
12. Microsoft Power Platform: wires the systems together.

Any tool doing the same job works; the method is the point. Full stack: https://pro.gtmvault.co/stack/operator/umar-farooq-adam

Where it breaks

1. No structural buying window. The technographic layer carried the heavy lift for this client because a supplier exit forced a move. Without an equivalent event, the score ranks fit, not timing.
2. A score that never reaches a rep. If the activity loop is not wired back from the sequencing tool, the list becomes another static spreadsheet and lead time does not move.
3. Treating derived estimates as facts. Umar labels budget and storage estimates arbitrary; weighting them heavily lets guesses outrank observed signals.
4. Downstream executors without context. When partners or reps receive accounts without the brief and the signal, precision upstream is wasted at the call.
5. Underestimating setup. The first two months of this engagement went to setup alone (SMYS 5, 1:53).