How the research engine decides who to contact
Before a sequence is written, research decides which accounts are worth contacting and why now. It runs as four connected stages, and people stay in the loop at every point that reaches a prospect.
The short answer
The research engine monitors real-world signals for your industry, enriches flagged accounts with verified contact and firmographic data, scores each one against your ICP, and hands the result to messaging that references the actual reason for contact. It is not a static purchased list, and it keeps running.
This is the first half of signal-based outbound: reaching an account because something just happened that makes your offer relevant, instead of working a cold list in bulk. The second half, channels and ground rules, is on the Outbound overview. The managed process that sits around it is described on how managed outbound works.
The pipeline
Four stages, one continuous process
Each stage feeds the next, and the result lands in the same contact record in DeliHub.
- 1
Stage 1
Signal monitoring
Watch the events that mean an account is worth reaching now.
- 2
Stage 2
Data enrichment
Add verified firmographics and decision-maker contacts.
- 3
Stage 3
Scoring
Rank accounts against your ICP so effort goes where it fits.
- 4
Stage 4
Messaging handoff
Pass the signal into copy that states the real reason for contact.
What counts as a signal
A signal is a timely event that suggests an account may need what you sell right now. There are two kinds, first-party and market, and the ICP workshop decides which ones matter for you.
Two kinds of signal
- First-party signals. DeliChat identifies the company behind a website visit and, where the law and the visitor’s consent allow, the person. Page views, return visits, opens, clicks and replies feed that person’s intent score, and replies are qualified before a meeting is booked. See visitor identification.
- Market signals. Role changes, funding rounds, hiring and other recent events pulled by our research before a sequence is written.
The intent score tracks one person’s behaviour. The ICP fit score is separate and tracks how closely an account matches your ideal customer profile.
Signals are monitored on an ongoing basis rather than checked once. We do not claim a specific reply-rate or meeting uplift from signal-based outreach. What it changes is relevance, timing and the number of irrelevant touches.
Account record
What a signal-led account looks like
Both kinds of signal on one record, with the account’s ICP fit score.
Northbridge Advisory
Management consultancy, generic example
- Funding round announcedMarket
- New operations lead hiredMarket
- Visited the services page twiceFirst-party
- Opened the first emailFirst-party
Next step: a sequence is drafted around the trigger and waits for a person to approve it. First-party signals appear only where the visitor’s consent allows.
The four stages in detail
Each stage has one job. Together they turn a broad market into a short list of accounts with a reason to be contacted.
Signal monitoring
We define, per client, which real-world events indicate an account is worth reaching now. Depending on your ICP that can include funding events, leadership changes, hiring surges, technology adoption, filings or expansion announcements. They are watched continuously, not checked once.Data enrichment
Flagged accounts are enriched with verified firmographic detail and decision-maker contact information. Enrichment draws on several sources rather than one static list, because any single source is incomplete, and every contact is verified before use.Scoring and prioritisation
Each enriched account gets an ICP fit score against your ideal customer profile: industry fit, company size, signal strength and likely budget. Outreach concentrates on the accounts that fit best instead of being spread evenly across a broad list.Handoff to messaging
The signal and enrichment data feed message generation, so the outreach cites the actual reason an account was contacted rather than a mail-merge. How that copy is written and approved is covered in AI personalisation.
A worked example
One client’s segment shows how a defined trigger turns into a first wave of outreach. It is an illustration of the method, not a benchmark.
A management consultancy client sells to private-equity-backed portfolio companies, among other segments. Here is how the method applies to a segment like that: the ICP workshop agrees that a recent ownership or funding event is the strongest trigger, monitoring watches for it, enrichment confirms company profile and the right buyer, each match is scored, and the highest-scoring accounts form the first wave of outreach.
Illustrative walk-through
From trigger to first wave
The method in four steps. The steps are illustrative, and the real segments are in the case study.
- 1Workshop
Agree the trigger
A recent funding event
- 2Ongoing
Monitor
Watch for matching events
- 3Per match
Enrich and verify
Profile and contacts
- 4First wave
Score and prioritise
Best fit goes first
What this is not
The engine speeds up research. It does not replace the people who write, approve and qualify.
- Not a static purchased list. Signals are monitored continuously and data is refreshed and verified, not scraped once and reused for months.
- Not one black box. It is a pipeline of coordinated stages, which we would rather explain accurately than oversell.
- Not fully autonomous. A person approves every sequence before it sends, and replies are qualified by people.
- Not identical for every client. The signals are set in the ICP workshop for your market, not applied as one formula to every industry.
Who does what
Software does the heavy lifting, people make the calls
Blocks are stages, not hours. Colour shows who owns each stage.
- 01Signal monitoring
- 02Enrichment
- 03Scoring
- 04Copy drafts
- 05Sequence approval
- 06Reply qualification
- 07Weekly review
AI and software
Monitoring, enrichment and scoring run continuously.
AI plus a human writer
AI prepares first drafts, a writer refines them.
A person
Approves every sequence, qualifies replies, reviews each week.
Questions about the research process
Straight answers on data, signals and where people stay involved.
Is this the same process for every client?
The stages are the same: signal monitoring, enrichment, scoring and a human-approved handoff to messaging. Which signals matter changes for every client. A software company may care about funding events and hiring, while an accounting firm may care about new business registrations and growth signals. The ICP workshop decides which signals we watch for you.
Do you use third-party data providers?
Yes. No single data source is complete or accurate on its own, so we combine licensed B2B contact data with public and firmographic signals to keep prospect profiles current. We do not publish the specific providers, because that is operational detail we manage on your behalf.
How do you avoid contacting bad or outdated data?
Every contact is passed through email verification before it enters a sequence, and enrichment is refreshed rather than reused from an ageing list. Bounced or invalid contacts are removed from future sends.
What is a signal in practice?
A signal is a timely event that suggests an account may need what you sell right now, such as a funding round, a leadership hire, a hiring push, a technology change or a new registration, depending on your industry. Signals are what separate a relevant message from a generic one.
Does the AI scoring replace human judgement?
No. Scoring narrows a large list to the accounts most worth reaching. Human writers write the sequences, a person approves every sequence before it sends, and replies are qualified against your ICP, timing, budget and role before a meeting is booked.
Does the research engine use website visitor data?
Where DeliChat is running on your site and the visitor’s consent allows, website visits can feed in as a first-party signal. Identification is consent-aware and depends on the visitor’s jurisdiction. See the visitor identification documentation for how that works.
Can I see why a specific meeting was booked?
Yes. A qualified meeting arrives with a short brief on the account and the conversation that led to it, and the contact record in DeliHub holds the history, so you can see what prompted the outreach.
Keep reading: Outbound overview, deliverability, personalisation and the B2B lead generation service. For sector detail see industries, and for packages and prices see pricing page. How we handle prospect data is in data privacy.