Is your AI stack creating activity or building pipeline? An 8-point checklist
The short answer
Before adding any new AI tools, check whether your current stack is creating activity or building pipeline. Test each part against one question about the results it leads to: data enrichment, intent signals, outbound, personalisation, call intelligence, CRM reporting, account strategy and sales content. Then connect the tools that work.
Key takeaways
- The average cold email reply rate is 3.43%, while the best-performing senders get more than 10% by favouring precision and personalised, problem-focused messaging over volume (Instantly, 2026).
- Most AI tools increase activity. A growing pipeline only shows up when you ask what each tool's output resulted in.
- The biggest gains usually come from connecting the tools you already have, so intelligence carries across from one stage to the next.
Your team is using AI tools. Spend is increasing and usage limits keep creeping higher, but it isn't clear what any of it is actually delivering for the business. The tools feel disconnected, and when the data or context changes, there's no clear process for making sure the models are updated and socialised accordingly.
What I see most often is tools being used in isolation, each doing its own job well according to its instructions and objectives. The research doesn't always reach the call prep, the content doesn't feed the nurture sequence, and what you learn in client conversations never makes it back upstream to marketing. It's the same problem commercial teams have always had: siloed information, missed handovers and intelligence that doesn't travel. I wrote about this stage in Which commercial problem do you fix first?
This checklist covers the eight areas I look at in any commercial AI stack. For each one, there's the activity trap to watch for and one question that tells you whether it's creating pipeline.
Activity signals and pipeline signals
- Emails and sequences sent
- Records enriched
- Licences in use
- Assets and posts produced
- Calls recorded
- Replies from the accounts you targeted
- Meetings booked with the right people (scored for likelihood)
- Opportunities created and progressed within the expected timeframe
- Win rate and time to close
- Revenue from existing customers
Data enrichment and prospecting
Tools that build or enrich your account and contact lists with firmographic and contact data.
The activity trap: It's easy to enrich thousands of records and feel like the team is well prepped, while the accounts that actually fit your ideal customer profile get lost in the volume.
The pipeline test: What share of the accounts you enriched last quarter became real conversations?
Intent and signal detection
Tools that flag potential buying signals: patterns of behaviour such as hiring decisions and investment announcements, or website visits with onsite behaviour data.
The activity trap: Signals are monitored, shared or reported in a dashboard that nobody is responsible for actioning, so warm signals get missed.
The pipeline test: When a signal is triggered, does someone act on it within a day, and do those accounts convert better than the rest? Do you have examples of when signals have been missed?
AI-assisted outbound and sequencing
Automated or AI-drafted sequences, cadences and outreach scheduling.
The activity trap: Sequences make it easy to double the volume of outreach, and reply rates fall as buyers learn to ignore overly generic or scattered approaches.
The pipeline test: As your sending volume has gone up, has your reply rate dropped, been maintained or improved?
Cold email reply rates in 2025
Share of cold emails that received a reply, by sender group
AI-personalised messaging
Tools that draft individually tailored copy at scale, beyond a single mail-merge template email.
The activity trap: Mail merge is seen as personalisation. A first name, a company name and a line about their latest LinkedIn post say nothing specific to the buyer.
The pipeline test: Could this message be sent to a competitor's prospect with just the name changed?
Conversation intelligence and call coaching
Call recording, transcription and coaching software used on live sales conversations.
The activity trap: Hours of calls get transcribed and summarised, but the insight never gets discussed in a 1-to-1, shapes a proposal or gets included in a marketing brief.
The pipeline test: In the last month, has anything that was extracted from a call transcript changed how the team sells or been fed into marketing content?
From experience
I spent several months running a full agentic system across my own commercial function, initially customising off-the-shelf plugins and skills and then building my own: research, outreach, call prep, pipeline analysis, marketing content, follow-up and brand assets. At first I followed a task-by-task automation process. The research ended up being superficial, the content was thin, and insights were buried in the discovery stage and didn't get carried through to the follow-up and proposal. So I paused and rebuilt it around the customer journey, designed to pass intelligence forward and built around shared objectives that served the customer first, instead of automations and agents working in silos. Rather than a new tool, the fix was building what went around the tools, resulting in a joined-up operating system.
CRM and pipeline reporting
A CRM or reporting tool the team actually keeps up to date.
The activity trap: The CRM is licensed but treated as a burden that takes away from more productive use of time. Data is entered sporadically, so pipeline reports get rebuilt by hand every week and are out of date the moment they're completed.
The pipeline test: Would you make a forecast decision from your CRM today without checking with the team first?
Account strategy and stakeholder tracking
A live view of who the stakeholders are on each deal, the strategic plan to win it and how well you're connected.
The activity trap: Stakeholder notes sit in personal files or, even worse, in someone's own shadow AI system. Deals rely on a single contact, and nobody is tracking progress along the sales cycle.
The pipeline test: For your five biggest open deals, can the team name every stakeholder and the roadmap to closing and onboarding the client?
Pitch and sales enablement content
Sales decks, proposals, one-pagers and campaign assets built systematically, instead of from scratch each time.
The activity trap: AI makes new content quick to produce, so the library grows while sales either can't find what they need or revert to, or rebuild, their own version.
The pipeline test: Which assets were used in the deals you won last quarter, how do you know, and who has access to them?
The 8-point checklist
Answer each pipeline test for your team as it is today: not yet, sometimes, or yes, consistently. Skip any area you don't have a tool for yet.
AI stack checklist: activity or pipeline?
A printable version of the eight areas and pipeline tests, with space to note the tools you use, who owns each one and what to fix first. A starting point to adapt to your own stack and sales process. 2 pages.
Get the PDF, free
Where to start
Map your current tools against your end-to-end commercial flow, from first contact to won deal to retained customer. Use the ACCR loop as a framework unless you already have one in place. Most teams find their tools clustered at the awareness end and almost nothing at retention, and this is often where the quickest value can be generated. Then look at the handovers between stages, because that's usually where the value is leaking, and build connections there before you add anything new. I set out how to do that in how to map your sales, marketing and BD workflow, and why the measures matter in the five gaps AI exposes in commercial teams.
Sources
- Instantly, Cold Email Benchmark Report 2026 (January 2026, covering 2025 data)
Where this fits in WISE: this is part of System orchestration. See how the full WISE system works.
Questions people also ask
How do you know if your AI sales tools are working?
Ask what each tool's output led to, instead of how much it produced: impact versus activity. Track replies from target accounts, meetings booked, opportunities created and win rate for the work each tool touches, and compare it with how things were before.
What should a commercial AI tech stack include?
Most cover eight areas: data enrichment, intent signals, outbound sequencing, personalised messaging, conversation intelligence, CRM and pipeline reporting, account strategy, and sales enablement content. What matters most is how well they share information with each other.
Should you add new AI tools or connect the ones you have?
Connect the ones you have first. Most of the value is lost in the handovers between tools and teams, so a new tool added to a disconnected stack usually adds more activity without more pipeline.
How do teams avoid ending up with too many AI tools?
Map your workflow before you buy anything, so each tool has a clear job at a specific stage. Give every tool an owner and one pipeline measure, and connect the ones you already have before adding more. How to map your sales, marketing and BD workflow shows you where to start.
About the author
Sarah L Morgan has led sales, marketing and BD teams for more than 20 years, from FTSE 100 media to £4m professional services firms. She now helps £5m to £30m businesses fix the people and process gaps AI exposes.
Seeing this in your team?
30 minutes on what's getting stuck. No pitch deck, no prep needed.
Book a 30-minute call