Orchestration, not automation: B2B leaders cut through the AI hype
AI is everywhere in pitches. It sounds impressive. And yet, three senior ABM leaders say what most teams need isn't more automation - it's better orchestration.
The message was blunt: AI is a tool. It accelerates smart strategy. It does not replace it.
AI is an enabler, not the strategy
Divya Handa, senior director of international marketing at Avalara, called out "a lot of AI-washing." Avalara has used agentic AI for two years across sales workflows, CX, and ABM. The lesson: meaningful lift comes from how you design the system around it.
"AI is an operational enabler. It helps with scoring, orchestration and surfacing early intent. But it does not replace human insight," Handa said.
Signal quality beats signal volume
Sophie-Louise Vevers, a senior demand programs manager, sees a flood of data - and weak decisions. A white paper download is treated like a buying signal. It isn't.
Real intent appears only when you layer multiple data points: first-party behavior, third-party intent, firmographic context, and historical activity. "One signal doesn't tell you anything. It's about painting the picture," she said.
From contacts to buying committees
Laura Winnan, EMEA integrated marketing manager at Zoom, urged teams to validate signals at the account level. "We look at accounts, not individual contacts. It's the pattern of behavior across the organization that matters."
AI helps move faster - Zoom's AI Companion included - but orchestration still depends on people choosing the right next action once a signal is confirmed.
Brand-to-demand and the 95:5 rule
Zoom operates on the 95:5 model: only 5% of your TAM is in-market now; 95% is out-of-market but still influenceable. The work is keeping your brand present so you're the default when intent appears.
If you want to see the research behind the idea, read the B2B Institute's explanation of the 95:5 rule.
Sales alignment makes or breaks it
Winnan was clear: if sales isn't in from the start, the whole system breaks. Don't dump data on reps. Give them the "why," the target accounts, and the next best actions.
Proof that orchestration pays
Avalara audited its entire data ecosystem - CRM hygiene, intent platforms, first-party behavior, and account development outreach. They found spend leaking into weak-fit accounts that rarely converted.
By refocusing on strong-fit accounts with meaningful signals and tightening coordination between data sources, Handa's team increased pipeline by 28% without increasing spend.
Where teams get stuck
- Treating a single action as intent (e.g., one download equals SQL).
- Automating tasks without defining the playbook behind them.
- Buying tools faster than they can be integrated or measured.
- Sending "intel" to sales without the context to act.
A practical playbook for AI-assisted orchestration
- Define your ICP and PQA rules: Agree on firmographic, technographic, and behavior thresholds that qualify an account - not just a contact.
- Layer your signals: Combine first-party engagement, third-party intent, historical activity, and persona fit. Score the pattern, not the click.
- Map signals to next actions: For each signal pattern, pre-define the sequence: who acts, what they do, and in what channel (ADR outreach, executive email, content syndication, event invite).
- Instrument the handoff: Push context to sales inside their workflow (CRM/SEP) with short briefs: "Here's the 'why', here's the talk track, here's the asset."
- Close the loop: Track outcomes by account and persona. Update scoring weights and plays every quarter based on what converts.
Workflow architecture that actually works
- Data hygiene: Clean CRM, dedupe, normalize firmographics.
- Signal engine: Unify web analytics, intent data, product usage (if applicable), and enrichment.
- Decision layer: AI/ML for scoring and routing; human review for high-value accounts.
- Activation: ABM platform, marketing automation, sales engagement - with shared playbooks.
- Feedback: Outcome data feeds back into the scoring model and playbook updates.
Tech stack sanity check
- List every tool and its single, provable job. Remove overlap.
- Measure time-to-signal, time-to-first-touch, and conversion by signal pattern.
- Adopt internal AI where it adds speed to judgment (e.g., summarizing account signals, generating first-draft talk tracks), not to replace it.
Upskill the team; don't just automate campaigns
Several firms are training internal models on proprietary data to spot lookalikes and shape pitches. Adoption is early. "We're quite immature as an organization in AI," one panellist said. The opportunity is designing workflows around AI so people can act intelligently, fast.
If you're formalizing team capability, explore a skills-focused path like an AI certification for marketing specialists.
On personalization at scale
Avalara rolled out custom ChatGPT instances to help reps personalize outreach. Uptake was mixed - a common pattern. Tools only help if the play is clear, the prompts are embedded in the workflow, and the outcome is measured.
- Standardize prompt templates by persona and industry.
- Auto-attach the top 3 relevant insights from the account's signal history.
- Score each touch on relevance and response to coach reps and refine prompts.
The takeaway
Automation moves hands. Orchestration moves markets. Build the system that turns signals into the next right action, keep sales at the center, and let AI accelerate the parts people shouldn't do by hand.
Do that, and the pipeline grows without the burn.
Your membership also unlocks:
