Most signal tools tell you a company is researching a topic. An IT spend increase is a different kind of evidence: a measured change in a budget line, with a date. Inside Claude, the Lusha connector runs the check as a batch, keeps the “no signal” rows so you see the whole list rather than only the winners, and separates customers from prospects so the same signal routes to a CSM or an AE. The run costs signal credits only; no contact is previewed or revealed until you decide the account is worth it. Every signal is sourced under Lusha’s data methodology and compliance framework.
FAQ
What counts as an IT spend increase?
A rise in the company’s estimated technology budget allocation compared with the prior reading. The signal carries a size, a date, and the budget range it was measured against. It does not break down which categories grew.
How far back does the signal go?
The prompt asks for 90 days, which returns up to three monthly readings. You can set startDate to any date in the last six months. Older than that returns nothing.
Why rank by consistency instead of size?
Because in the live run the five largest increases were the five least believable. Two consecutive readings in the same direction is harder to fake with a data refresh than one large jump. Size still breaks ties among single readings.
Why keep the accounts with no signal in the table?
Because “no signal” is information. If 11 of 25 target accounts show nothing, the list itself is the question, not the reps working it.
Does this spend credits?
Signal reads cost 1 credit per event returned. In the live run, 25 domains cost 17 credits. The prompt runs account_usage first at zero credits and stops under 50, so you never start a batch you cannot finish. No contact preview or reveal runs in this play.
Can I run this without a list?
Yes. Replace step 1 with prospecting_company_search using your ICP filters plus the itSpendIncrease signal. Validate filter values with prospecting_company_filters first; unrecognized values return empty results without an error.
