So you’ve cleaned your CRM. You’ve fixed your data quality issues. Your AI tools finally have something decent to work with.
Now what?
This is where most marketing teams stall out. They’ve done all the foundation work we covered in Parts 1 and 2, but they’re still not entirely sure what to do next. They’re asking ChatGPT to write emails instead of building systems that actually change how work gets done.
Let me tell you about a CMO I worked with last quarter. She’d spent six months getting her data house in order. Clean CRM, solid enrichment process, proper governance. The works. Then she asked me, “Okay, now how do we actually use this stuff to grow faster?”
Fair question. Because clean data doesn’t generate pipeline. Systems do.
👋 Hiii, it’s Kaylee Edmondson and welcome to Looped In, the newsletter exploring demand gen and growth frameworks in B2B SaaS. I write this newsletter every Sunday, and wildly, a few thousand of you read it each week. I’m grateful. If there’s ever anything in particular I can help explore, research, and write about, hit me up! I’d love to chat.
The workflow automation nobody talks about
Here’s what I see happening at most B2B companies right now: marketers are using AI to speed up tasks that probably shouldn’t exist in the first place.
A client came to me three months ago. Their SDR manager was spending 12 hours per week manually researching leads on LinkedIn, identifying the right contact at each account, enriching the data in spreadsheets, then routing leads to the right reps based on territories and availability. They even had two part-time VAs helping with research—$2,500 per month just to keep the process moving.
They wanted to use AI to make the workflow faster.
I asked them why the workflow existed in the first place.
Turns out, they had never set up an automated enrichment or routing system because they “didn’t trust automated data pulls” and worried routing rules would break. So they built a human-powered enrichment machine instead.
We rebuilt the entire workflow—Clay for enrichment and buying-committee identification, plus native CRM routing rules to handle assignment based on real-time rep capacity. It took about a week.
Now the whole thing runs automatically. It enriches every lead, finds the right stakeholders, and sends them to the right rep instantly.
That 12 hours per week (plus the VA bill)? Gone. The SDR manager now spends that time on coaching, strategy, and helping reps close more pipeline instead of copy/pasting data between tabs.
Some actual automation opportunities in your GTM motion:
Ad operations: Stop manually adjusting your LinkedIn Ads campaigns on a hunch (or multiple times per day). Modern tools like Fibbler automatically push your LinkedIn ad engagement data (impressions, clicks, etc.) into your CRM on a weekly cycle, letting you visualize account-level trends that no standard ads dashboard can easily surface. This means you’ll spot which target industries or accounts are responding, and you can tweak creative or budget each week based on real patterns instead of guesswork. For example, one team saw engagement spike after launching a storytelling-style video ad – a trend the weekly data made obvious – so they doubled down on that format to sustain the gains. In short, set clear goals and let the algorithms optimize in real time, using these insights to focus your energy on creative strategy rather than fiddling with bids. Here’s a step-by-step by Fibbler directly.
Content briefs: If you’re still manually crafting detailed content briefs for every SEO article, stop. AirOps provides AI-driven templates that turn your SEO research into structured, on-brand content outlines automatically. Feed it your target keywords and context, and it generates a ready-to-use brief complete with suggested headings, internal links, and tone-of-voice guidelines from your brand kit.
Voice of the customer: Stop digging through Slack threads or Gong call transcripts to find customer feedback. A tool like Glean indexes and searches across all those conversations, so anyone in Marketing, Product, or Sales can instantly pull up what customers are asking for or complaining about. Instead of spending hours piecing together scattered messages (and inevitably missing key insights), teams can query and get a consolidated view of every relevant Slack chat, support ticket, or sales call note on a topic. It even aggregates daily digests of customer feedback from Slack, emails, calls, etc., highlighting top pain points and quotes without you lifting a finger. 🤯
The pattern here? Automate the repetitive execution work so humans can focus on the strategic decisions that actually require judgment.
Also, not sponsored, btw. Genuinely just the tools I’m using across my client base right now.
Building decision loops that don’t suck
Ok, stick with me. Here’s where it gets interesting. Yes, we can all start automating more tasks with a proper foundation, sure. But what happens when we add decision loops where AI and humans each do what they’re exceptionally good at…
I use a modified version of the OODA loop framework, but work backwards from the business objective:
Start with the decision you need to make. Not the data you have. Not the tool you just bought. The actual decision. Example: “Which accounts should we target with paid ads this month?”
Define the actions that decision enables. If you target these accounts, what happens? Budget gets allocated, ads get served, sales gets notified. Map the whole chain.
Identify what analysis you need to make that decision. Historical conversion data, intent signals, engagement patterns, budget constraints, competitive activity. Be specific.
Work backwards to the required data. Only now do you figure out what data you actually need. Most teams do this in reverse and end up with dashboards full of metrics that don’t drive decisions.
Build the loop. Humans set the strategy and constraints. AI processes the data and recommends actions. Humans review, refine, and approve. AI executes. Results feed back into the model. Repeat.
Here’s how this looks in practice with one of my clients:
They wanted to improve lead routing. Sales was complaining that marketing was sending them garbage. Marketing was frustrated that sales wasn’t following up fast enough. Classic problem.
We built a decision loop:
Human input: Sales and marketing jointly defined what “good fit” actually means. ICP criteria, budget indicators, buying signals. They also set response time SLAs.
AI execution: The system scores every lead using those criteria, routes high-fit, high-intent leads to the right rep based on territory and availability, and flags edge cases for human review.
Human refinement: Every week, they review a sample of leads and routing decisions. When sales marks a lead as “not qualified,” the system learns why. When a “low score” lead converts, we investigate what the model missed.
Three months in, lead acceptance rate went from 62% to 89%. Response time dropped from 8 hours to 23 minutes (during business hours). Sales and marketing finally agreed on something. Woo!
The key insight: humans train, AI executes, humans refine. It’s a continuous cycle, not a one-time setup.
The compliance stuff you can’t ignore
I need to talk about governance because I’ve seen too many companies build amazing automation systems and then get absolutely wrecked by compliance issues.
A client in Europe built this beautiful AI-powered personalization engine. Dynamic website content, customized email flows, predictive lead scoring. It was genuinely impressive.
Then their legal team found out they were using AI to make automated decisions about customer treatment without proper consent mechanisms or explanation capabilities. GDPR violation waiting to happen. They had to shut the whole thing down and rebuild with proper governance from day one.
Here’s what you do instead:
Consent management that works: Your CMP needs to integrate with your automation platforms. When someone opts out, that needs to propagate everywhere within 24 hours, ideally instantly. Set up automated purge workflows. Test them quarterly.
Audit trails for AI decisions: Every significant automated decision needs to be logged. What data was used, what the AI recommended, whether a human overrode it, what happened. You need to be able to explain why the system did what it did six months later.
Bias detection: Run regular audits on your AI outputs across different demographics. Are certain segments getting systematically deprioritized? Is your account scoring biased toward specific company types? Check the actual distribution of outcomes, not just what the model says it’s doing.
Human checkpoints for high-stakes decisions: Budget allocation over $10K? Strategic account assignment? Campaign strategy for a major launch? Those need human approval, not just automated execution. Build those gates into your workflows from the start.
The good news: most modern platforms have these capabilities built in. You just need to configure them instead of clicking through the setup wizard and calling it done.
Orchestration vs. point solutions (and why everyone gets this wrong)
Every CMO asks me the same question eventually: “Should we get one big platform or best-of-breed tools?”
Wrong question.
The right question is: “What decisions do we need to make, and what’s the simplest system that enables those decisions reliably?”
I worked with a Series B company last year. They had 23 different marketing tools. Twenty-three. Each one was “best in class” for its category. Their stack diagram looked like a bowl of spaghetti.
Their marketing ops person spent 60% of her time just keeping data synced between systems. Not optimizing campaigns. Not building new workflows. Just making sure that when someone filled out a form, that data eventually made its way to the right seven places without breaking.
We consolidated to an orchestration platform (HubSpot in their case, but the principle applies to any solid marketing automation platform) plus four specialized tools for things that genuinely needed best-of-breed solutions.
Time spent on integration maintenance dropped to maybe 10%. The rest went toward actually improving their GTM motion.
Here’s my framework for this decision:
Use an orchestration platform when: You have more than 10 tools creating data silos. Teams argue about whose numbers are correct. You have complex sales cycles with multiple touchpoints. You need to prove ROI across the full buyer journey. You want to scale without proportional headcount increases.
Stick with point solutions when: You have simple sales processes. You’re an early startup with limited budget. You have specific workflow needs that general platforms can’t handle. You have technical resources to manage integrations properly.
Use the hybrid approach when: You’re a mid-market or enterprise company with complex needs but specialized workflows. You need the unified data layer of an orchestration platform but also need depth in specific areas. This is where most B2B SaaS companies end up, honestly.
One pattern I see working really well: orchestration platform as your central hub, connected to 3-5 specialized tools for specific high-value workflows. Data flows through the central platform. Single source of truth. But you get depth where you need it.
Clay for enrichment and research. A proper CDP if you need sophisticated segmentation. A new-stack ABM tool (read: not 6sense) if you’re running coordinated sales and marketing plays across large enterprises. Whatever makes sense for your specific motion.
Every tool needs to justify its existence by enabling decisions you couldn’t make otherwise. If it’s just giving you “more visibility” or “better insights,” you probably don’t need it.
Don’t try to do all of this at once. I’ve seen companies spend $200K on fancy AI platforms and get zero value because they tried to implement everything simultaneously and got overwhelmed.
Start small. Prove value. Build momentum. Scale gradually.
Your data is clean. Your foundation is solid. Now go build the engine.
See ya next week,
Kaylee ✌
Wow, 'systems do' is so spot on. It's about designing inteligence into workflows.