Intent leaderboard

This is an internal prospecting screen for Stripe account executives. It answers one question: which accounts in Stripe’s ICP are showing a pattern of buying intent right now, rather than a single stray mention? The leaderboard lists accounts where two or more distinct signals fired inside a 90-day window (14 July to 12 October 2026), ranked by how many distinct signals fired and then by how recently. Every row carries chips for the signals that fired and the headline of its newest event; opening a row shows all of that account’s events with the date, the publication, the Stripe product the event points at, and a link to the source. Job-change events render the person who moved and the role they left. Accounts that fired exactly one signal sit behind a filter and are labelled as noise, and a “Strong (4+)” band picks out the 23 accounts with four or more distinct signals. A second view shows every signal’s fire rate measured against all 369 accounts scanned.

Everything on screen came from the Trayo API. The candidate set is one trayo_find_companies search in direct mode over the workspace’s own ICP — twelve industries from the taxonomy, a 200-20,000 headcount band, seven English-language headquarters countries — paged four times for 400 rows, of which 379 carried a website. Those were imported with trayo_import_accounts (minus Stripe itself and four junk listings) as 369 accounts, then run through two trayo_run_discovery waves at a 90-day lookback against all ten signals the workspace was onboarded with. Wave one covered 198 accounts and 65 of them cleared the two-signal bar; wave two covered the remaining 171 and took the total to 123. Discovery wrote 731 events, 609 of which fell inside the 90-day window — the other 122 were job changes dated by the month the role started, which sits before the window. 244 of the 369 accounts fired at least one signal. Intuit led with eight distinct signals across fourteen events; Davidson Hospitality Group and Western Union followed at seven.

Three honest notes. First, I set the target at 25 accounts and the first wave alone returned 65, so the second wave was not strictly needed — it was already running in parallel when the first settled, and it is included. At 123 of 369 the two-signal bar is a filter, not a shortlist, which is why the app says so and offers the four-signal band; growth-scaleup fired on 35.5% of accounts and finance-automation-hiring on 27.1%, and the signals view flags both as too broad to carry a shortlist on their own. Second, a filters-only Trayo search returns largest-first, so although the ICP band was set at 200-20,000 employees the accounts actually scanned run 7,523-19,928 — smaller ICP accounts are not represented. Third, “Push to Salesforce” and “Draft outreach” are stubs: no CRM is connected and no mail is sent; each confirms with the library’s preview toast saying what would have been sent. On branding, brandFromWorkspace gave #635BFF the primary role and read #0A2540 as an accent (chart series two onward) rather than as the shell, so the top bar wears a deep step derived from the violet rather than Stripe’s navy. The library also reported that #635BFF falls under 4.5:1 as text on a light page and substituted #5446EA for links and accent text, and that with no background colour supplied the page takes a soft violet tint. Those are the library’s own decisions and were left as it made them.