A signal-driven GTM motion starts outreach from a real change at an account, such as a new hire, a funding round, a job post or a new tool in their stack, rather than from a static list or a form fill. This playbook covers five plays: research, outbound, personalisation, account planning and follow-up. Each has three levels: the old manual way, what one rep can do today with AI, and the automated version a GTM engineer builds.
I spent five years at Salesforce, first as a BDR and then as an Account Executive. Almost everything I did then was manual: hours in annual reports, guesswork account plans, generic templates and days spent building lists. The same patterns still show up in most teams we meet. They automate outreach before fixing enrichment, buy AI before defining their ICP, and chase signals without designing the system that acts on them.
The fix is not one tool. It is a way of working, and you can adopt it one play at a time.
Play 1: Research that is better, not just faster
Level 1, manual. Skim annual reports and earnings calls and hope you catch the point that matters. Thirty to sixty minutes a document.
Level 2, AI today. Use a deep research tool such as Gemini or ChatGPT with a precise brief. For example: “Summarise this company’s latest annual report. List the CEO’s top three strategic challenges, the main priorities leadership mentioned and any risks they flagged.” Review the output, then keep everything for an opportunity in one project so the context builds up.
Level 3, automated. Run the same prompt as an AI column in Clay across every target account. When the summary picks up a phrase that matters to you, such as “expanding into Europe” or “hiring a customer experience leader”, the account is scored higher, moved to a priority list and an outreach draft is prepared.
Play 2: Outbound that earns attention
Outbound is not dead. Outbound built on static lists, weak signals and generic copy is.
Level 1, manual. Download a thousand accounts that match on industry and headcount, write one semi-custom sequence and send it to all of them. No signal, no timing, no reply.
Level 2, AI today. Pick ten to twenty accounts, find two or three strong signals for each, and write one properly personalised message a day with AI helping on the first draft.
Level 3, automated. One client was targeting Series B software companies expanding into the UK and Ireland. They had a strong ICP but no way to act on change in real time. The workflow we built:
- Tracked their 300 top accounts for new VP of Sales or Head of GTM hires.
- Tagged accounts that announced funding or started hiring for enablement roles.
- Drafted outreach aligned to the specific signal, for the AE to review.
- Pushed approved messages into the sending tool and assigned the AE in HubSpot.
Now they can see which signal led to which reply and which meeting. That feedback loop is the point.
Play 3: Personalisation that reflects real context
“Congrats on the funding” is not personalisation any more. Relevance comes from the buyer’s actual situation.
Level 2. Have Clay watch LinkedIn posts, company news and press releases for the signals you care about, flag them as they appear and draft a message that references the trigger. A person reviews it before it goes.
Level 3. Monitor several sources on a daily schedule, push each signal into the CRM, Slack or a database with the context attached, segment by trigger type and generate the whole short sequence for that context.
Another client monitored more than 800 target accounts for job posts mentioning European expansion, for French or German being added to their website, and for new heads of customer experience or RevOps. Accounts were scored and tagged as signals arrived, a tailored sequence was generated, and the AE got a Slack message with the context. Nobody had to open the CRM to find the opportunity.
Play 4: Account planning from signals, not spreadsheets
An account plan should not live in a deck that is updated once a quarter.
Level 2. Bring funding, hiring, tech stack, leadership moves and website changes into one Clay table, set filters that reflect your ICP and sort accounts into tiers. Sync the result to the CRM or a shared sheet you use in pipeline reviews.
Level 3. Build a scoring model that updates itself as data refreshes. An illustrative example:
- Raised funding in the last six months: 20 points.
- Uses the CRM and call tools you integrate with: 10 points.
- Hiring for operations or customer experience: 15 points.
- Leadership posting about retention: 5 points.
Set tier thresholds, then let changes in tier trigger the next step: create or update the CRM record, add high scorers to outbound, or give the AE a task. The weights above are an example only. Yours should come from your own closed-won data.
Play 5: Follow-up that is structured, not scrambled
Follow-up is where deals are won or lost, and it is still where most teams are slowest.
Level 2. Call recorders such as Gong, Chorus or Otter transcribe and summarise the call. Better, but you still turn the notes into actions by hand.
Level 3. Clay can pull call transcripts in directly. AI extracts the themes, objections and buying signals, the account is enriched with fresh data, and the workflow drafts the follow-up, updates the CRM, alerts the rep in Slack and creates a task if a new stakeholder or competitor was mentioned.
- Competitor mentioned: draft a follow-up with the relevant battle card.
- Feature requested: alert the product team.
- New stakeholder mentioned: enrich their record and add them to the opportunity.
Where to start
Pick one play and one signal. If your CRM data is patchy, start with the foundation first, because every play above depends on it. Our 30-day foundation plan shows the order we use.