Privacy: companies masked by letter, no people named, no article titles or URLs. Examples pulled live via the Lusha connector, October 31, 2026. No percentages or sample statistics appear in this piece. Data counts use 515M+ and 29M+ tokens if added.
You see a news signal on a target account, the article ran this week, and you open with “Congrats on the launch.” The launch was in 2023. The article was a roundup that mentioned it. That is the one way a news signal hurts you, and it is avoidable, because every news signal Lusha returns in Claude carries two dates: when the article was published, and when the event it describes took place.
This guide is about the second date. Three rules, four examples from a live pull, and the prompt wording that makes Claude apply the rules for you.
Companies in the examples are masked by letter. No people are named. Examples pulled live via the Lusha connector, October 31, 2026.
The two dates on every news signal
| Field | What it tells you | What it does not tell you |
|---|---|---|
| Published | When the article went live. This is the date that puts the item inside your search window. | Whether the thing it describes is new. |
| Event date | When the event happened: a funding round’s close, a launch day, an executive’s start. | Nothing more. When the article does not say, the field is empty. |
They differ for ordinary reasons. Press releases restate history (“the company, which raised a Series F in 2021, today announced”). Roundups and comparison pages republish old facts under new dates. Analyst coverage of a quarter can run weeks after the quarter. None of that is wrong as journalism. It only becomes a problem when a rep treats the publish date as the event date.
The calculated signals (headcount, hiring surge, IT spend, new job posts) carry one date each, because Lusha computes them from the underlying data. The two date problem is specific to news.
Three rules for reading news signals
- The event date decides, inside your window. If you work a 90 day window, a news signal counts when its event date falls inside those 90 days. The publish date is how the item got into the list, not why you call.
- One article is one event. A single press release can produce several signals, sometimes under different signal types (a launch as product activity, the same launch as market intelligence). Group by article before you count momentum. Four signals from one release is one piece of news.
- Empty event date means verify, then call. When the article does not state when something happened, the field comes back empty. Treat that as a lead to check, not a trigger. Open the article, find the date, then decide.
Four examples from a live pull
| Company | Signal | Published | Event date | What the date field caught |
|---|---|---|---|---|
| T. (education software) | Product launch | October 2026 | April 2023 | A roundup republished a feature from three years ago. Rule 1: outside the window, no call. |
| J. (identity software) | Funding round | September 2026 | July 2021 | A revenue summary page restated an old Series F. Rule 1: outside the window. |
| W. (collaboration software) | Product launch, four signals | October 2026 | October 2026 | One conference announcement, four signals across two types. Rule 2: one event, one opening line. |
| S. (scientific software) | Partnership | August 2026 | August 2026 | Dated inside the window, a named enterprise customer, one article. This is the one to call about. |
The fourth row is the point. When the event date sits inside the window and the article is a first mention, the signal is doing its job: a dated reason to call, with the context in the article highlight ready for the opening line.
The prompt that applies the rules for you
Say “event date”, not “recent news”. Claude passes the window to the connector and filters the results on the field that counts.
Using Lusha, pull the news signals for [company domain] from the last 90 days.
For each item show the signal type, the publish date, the event date and a one-line summary.
Then apply three rules: 1) keep only items whose event date falls inside the last 90 days, 2)
group items that share an article URL and count each group as one event, 3) list items with no
event date separately under "verify first". End with the single best reason to call this week,
with its event date.What comes back: a short table of dated events, a “verify first” list, and one opening line with a date on it.
Credits are charged per signal item returned and shown in the response before you decide what to do next.
FAQ
Why does a news signal show an event from years ago?
Because the article was published recently and mentions the older event. The publish date put it in your window. The event date tells you it is history. Use the event date.
Why do I get several signals for one press release?
One article can describe more than one thing (a launch and a partnership, say), and the same event can be classified under more than one signal type. Group by article before you count.
What does an empty event date mean?
The article did not state when the event happened. Verify before you use it as an opener.
Do calculated signals have the same issue?
No. Headcount, hiring surge, IT spend and job post signals carry one computed date each.
Does filtering by event date cost extra credits?
No. Credits are charged per signal item returned. The filtering happens on what came back. Current rates are on the pricing page.
Related plays
- Get a single-account signal brief before the call
- Build a weekly signal digest on your saved territory
- Find accounts with stacked buying signals
- Write a cold sequence from a trigger event
More on every signal type: Lusha Signals hub.
