I took over a client's HubSpot account last year. Sixty-three workflows in the automation list. More than half were paused, producing errors, or doing nothing I could identify with any confidence. The team had been adding them reactively for three years: someone asked for a notification, someone else wanted a tag on certain contacts, a third person said "we should automate that" in a planning meeting. Nobody had built anything deliberately.
The first thing I did was pause everything non-essential and audit what remained. Then I rebuilt from a clean foundation. What followed here is that foundation, the four workflows I now set up on every HubSpot instance before I add anything else.
Why HubSpot automation gets cluttered so fast
HubSpot makes it very easy to add a new workflow. That is both its strength and its primary trap. When something is easy to add, teams add it without a system, and "one more workflow" becomes the default response to every friction point. Three years later you have 60 automations, many of which contradict each other or trigger on conditions that no longer exist in the pipeline.
The fix is not more discipline. It is starting with a clear mental model of what automation is actually for in a pipeline context, and building from that model rather than bolting things on.
The mental model: deal stages as state transitions
Think of your pipeline as a state machine. Every deal lives in exactly one stage at any moment. When a deal moves from one stage to the next, something should happen. That "something" is what your automation handles.
There are three types of events worth automating around each stage. Entry events trigger when a deal arrives in a stage: assign someone, create a task, send a notification. Stall events trigger when a deal has been in a stage too long without moving: alert the owner, notify the manager. Exit events trigger when a deal leaves a stage: log what happened, clean up properties, hand off to the next person.
Most teams set up entry events reasonably well. They almost never set up stall events. That gap is exactly where deals die quietly. A rep who stops working a deal does not usually make a conscious decision. The deal simply becomes background noise, and nobody knows until the quarter ends and the number is not there.
The four workflows that matter first
These are not hypothetical best-practice suggestions. These are the actual automations I rebuild or verify on every HubSpot account I touch.
1. New deal assignment and first-touch task
Trigger: Deal is created in the pipeline (or moves to the first active stage, depending on how the pipeline is structured).
Actions: Assign the deal to an owner based on whatever logic your team uses (round-robin, territory, source). Create a task due in 24 hours labeled something like "First outreach: call or email." If your team uses Slack, send a short internal notification to the assigned rep.
Why it matters: New deals have the highest conversion probability and the shortest window. A 24-hour gap between a deal entering the pipeline and the first rep contact is not just a missed opportunity, it is often the whole deal. This workflow makes the first touch the path of least resistance rather than something that depends on the rep remembering.
2. Stage stall detection
Trigger: A deal has been in the same stage for longer than a defined threshold. The threshold should vary by stage: appointment scheduled might be 3 days, proposal sent might be 7, negotiation might be 12. Set these based on your actual historical close data, not a guess.
Actions: Send an email or notification to the deal owner. If the deal is still stuck after double the threshold, include the manager in the next notification.
Why it matters: Most lost deals were not lost in a single moment. They got quiet. The rep had good intentions and then other things filled the day. Stall detection makes the silence visible before it becomes permanent.
3. Closed-won handoff
Trigger: Deal stage moves to Closed Won.
Actions: Create an "Onboarding kickoff" task and assign it to whoever owns delivery. Set a contact property for customer_since with today's date. If your team uses HubSpot for customer communication, optionally enroll the contact in an onboarding sequence.
Why it matters: The handoff between winning a deal and delivering on it is where first impressions collapse. Sales moves on. The customer is waiting to hear what happens next. This workflow removes the dependency on someone remembering to make that introduction, and it creates a clean record of when the relationship started.
4. Closed-lost reason capture
Trigger: Deal stage moves to Closed Lost.
Actions: Check whether the "Loss reason" property is populated. If not, set the deal back to a "Needs closure update" stage and create a task for the owner to complete it before it can close. Once the reason is captured, schedule a low-priority "reconnect" task for 90 days out and add a "past-prospect" tag to the contact for segmentation.
Why it matters: Without forcing the loss-reason field, you get a graveyard of dead deals with no data. After roughly six months of clean capture, loss reasons tell you more about where your pipeline leaks than any pipeline report will. The 90-day task costs nothing and occasionally resurrects a deal that was not lost permanently, just lost for now.
What to hold off on
The temptation once these four are running is to keep adding. Resist it for at least 60 days. There are two specific traps worth naming.
Do not automate your entire nurture sequence before testing it manually. Let a human send the first 20 or 30 follow-up emails and actually read the replies. Automation spreads whatever is in the sequence at scale. If the messaging is wrong, you want to find that out with a few dozen sends, not a few hundred.
Do not build multi-branch workflows for edge cases you have not actually seen. "What if the deal source is X and the contact type is Y and the owner is Z" logic is seductive to build and painful to maintain. Every branch is another thing that can break silently. Let the pipeline run for a while, watch what actually happens, and automate the real patterns, not the hypothetical ones.
The goal is a pipeline where the automation handles the administrative layer: assignment, reminders, stage logging, and data capture. The judgment layer, reading a situation and deciding what to say, stays with the person who owns the deal. These four workflows create that separation cleanly. Everything else builds on top of them.
Sources
Questions about setting this up for your team? Get in touch.
Comments