Key takeaways
- Personalise the reason for writing, not the greeting.
- A vague compliment reads as more robotic than an honest segment message.
- Tier your list so research time goes to the accounts worth it.
- Let AI research and draft, then cut every line that would fit another company.
What makes outreach sound robotic
Outreach sounds robotic when the personal parts are obviously bolted on. Readers notice the seam between the inserted line and the template within a sentence or two.
- Compliments that could apply to anyone, such as "I love what you are doing at your company".
- A first line about the prospect followed by a pitch that ignores it.
- Phrases nobody says aloud, such as "I hope this email finds you well".
- Every message the same length, with the same structure and the same ask.
- Merge fields that break, such as a company name in capitals or a job title copied from a database.
What does real personalisation look like
Real personalisation links something specific about the prospect to a reason to talk. It could only have been sent to one company.
Compare two openings to the head of operations at Larchwood, an invented furniture retailer. Robotic: "I saw Larchwood is growing fast and wanted to reach out." Real: "Your job advert for three warehouse planners mentions moving to next-day delivery; that usually puts pressure on returns handling, which is what we work on." The second gives the reader a reason to reply.
How do you scale personalisation without writing every line
Scale it by tiering your list and spending research time where the value is. Write fully personal messages to your best-fit accounts, give the middle tier a personal first line on top of a segment template, and send the rest a well-written segment message with no fake personal touch.
Segment templates do much of the work. A message written for heads of operations at mid-sized retailers is relevant before any personal line is added.
How can AI help without making it worse
AI helps when it does the research and the first draft and a person checks the result. Give the assistant your positioning, your voice and the trigger you found, and ask for a short draft.
Then read it as the prospect would. Delete any compliment, any claim you cannot back up and any sentence that would fit another company just as well.
How to personalise outreach at scale
Use the same five steps for every batch.
- Split your list into tiers by fit, and decide how much research each tier gets.
- Write one segment template per audience, in plain words and short enough to read on a phone.
- Find one real trigger for each top-tier prospect: a hire, a launch, a post or a change in their market.
- Write one sentence linking that trigger to the problem you solve, and put it first.
- Read a sample aloud before sending, and cut anything you would not say on a call.
Where Pebbles Ai fits
- CapabilityProspecting & customer recordsLeads found in Pebbles Ai land as CRM contacts (in Beta), each with personalisation angles from their public LinkedIn posts, falling back to their role and company.
- CapabilityResearch & analysisAsk an assistant to research a prospect's company from its public pages, with numbered citations, before you write the trigger line.
- CapabilityOutreach contentThe campaign builder writes email and LinkedIn pieces with a personality for each, and emails are saved as drafts in your connected mailbox so you read them before sending.
Common questions
How long should a personalised cold email be?
Short enough to read in one glance on a phone, usually a few sentences. Length rarely helps a first message; relevance does.
Is it dishonest to use templates?
No. A template is a starting structure. It becomes a problem only when it pretends to be personal, for example with a vague compliment that was clearly inserted.
Should every prospect get a personal first line?
Only where you have something real to say. A strong segment message beats a forced personal line.
About this page. Written by the Pebbles Ai team for people who run sales, marketing and growth. Last updated .
