M. Valera
GTM / outbound / judgment

outbound with AI: decide who not to contact yet

Marcos Valera ·

before you automate a message, there is a more important decision: does it make sense to contact this company now?

an account looking like your ideal customer does not mean it has the problem you solve. raising a round does not mean it is ready to buy from you either. there is work to do between the signal and the message.

I talked about this in my session with AI GTM Lab. you can watch the full session here. AI helps you research and build hypotheses. a person still decides who receives a message.

1. start with something you can check

a signal is something you can observe: a person's post, an event they attended, a funding announcement or a job opening relevant to what you sell.

save the source and the date. if you cannot show where you saw it, do not treat it as a confirmed signal.

write down only what you know. a company posting a job does not mean it has an approved budget for your product.

2. separate the signal from the fit

your ICP (ideal customer profile) tells you who you can help. a signal gives you context about timing. they are different things, so look at them separately:

an eye-catching signal does not fix a poor fit. and a good fit does not mean you have to write today.

3. a hypothesis is not a fact

the question is: why might this company need this now?

write an answer connecting what you observed to a possible problem. separate what is confirmed from what you are assuming. until you check it, it is a hypothesis.

look for clues in who they are hiring, what they post about and what their competitors are doing. none of that, on its own, confirms an intention to buy. if AI prepares the hypothesis, ask it for the evidence and for something that would disprove it too.

if you cannot explain why they would use your product and why now, do not contact them. do not even draft the message yet.

4. contact, wait or discard

this template is my adaptation of the process, to use as a worksheet:

field what to write down
signal what you observed, without interpreting it
source and date the original link and when it was published or you saw it
fit why the company and the person make sense
hypothesis what problem might matter to them and why now
validation what is confirmed, what is missing and what would disprove the hypothesis
decision contact, wait or discard, and why

contact: there is a fit and a specific reason to talk now. that does not mean they will buy.

wait: there is a fit, but you do not yet have evidence about timing. write down what would need to happen for you to revisit the account.

discard: there is no fit, no clear reason or the evidence contradicts your hypothesis. you do not need to invent another angle to keep the account in the sequence.

5. the message comes last

short, relevant and easy to answer. for example, a question someone can answer by choosing A or B.

use only what you have checked. frame the hypothesis as a question, not as if you knew their problems from the inside. do not promise savings or results you cannot support. and do not tie yourself to one channel: LinkedIn, phone and email.

before sending, a person reviews the source, the fit, the hypothesis and the text, then approves, edits or rejects it. if the hypothesis is wrong, polishing the copy will not fix it.

the process first, then the tool

decide what you want to observe and how you will judge it. then choose what to use for research or automation. if you want to know how I approach it, here is how I work on AI GTM.

sometimes the useful output is not a list of messages. it is a shorter list of accounts with a clear reason to talk (and another list of accounts you are not going to contact yet).

for me, outbound is not about sending more messages. it is about knowing why each account you approach should talk to you.

if you want me to talk about this with your team or at an event, here are my talks and sessions.

source: adapted from my session "Marcos Valera: Decide Who Not to Contact Yet", AI GTM Lab.