Find where the manual hours go, fix the data those tasks depend on, then automate the rule-based workflows with tools your team owns. Add AI last, for work rules cannot do, once the data it reads can be trusted.
Key takeaways
- Measure before you automate. List each recurring manual task and the hours it takes, so you have a baseline to prove the saving.
- Much manual work is a data problem. People keep their own sheets because they do not trust the CRM, and automating that work only spreads the errors faster.
- Automate rules, keep judgment with people. If a workflow cannot be written as rules on one page, it needs design before it needs software.
Find where the manual work lives
Most teams know they have a manual work problem, but few can name the five tasks behind most of it. Before you buy a tool, list every recurring task your revenue team does by hand, such as building the weekly forecast in a spreadsheet, routing leads, cleaning duplicates, assembling board reports or calculating commissions. Write down who does each one and how many hours it takes. That total is your baseline.
Then sort the list by cause. Some work exists because a process was never built. Some exists because two systems disagree and someone reconciles them every Monday. Some exists because nobody trusts the CRM, so people keep their own sheets. The last is the most common, and automating it only moves the bad data faster. A week or two of interviews with reps and the CRM owner is usually enough to see the pattern.
Fix the data the work depends on
Agree on one definition for each core field, such as what counts as an MQL, an opportunity stage or a renewal. Remove duplicates, set required fields and name an owner for each object. This is where reporting stops being a weekly chore: one client saved over 40 hours a month on reporting once lifecycle stages and lead sources meant something, and another fixed attribution so budget calls stopped resting on bad data.
Skip this and automation fails quietly. A routing rule sends the best accounts to the wrong rep, and a report looks precise while being wrong.
Automate the rule-based workflows
Good candidates are lead routing and scoring, quote and contract steps, renewal tracking, territory assignment and commission calculations. Write each as a plain rule set first: if this happens, then that happens, owned by this role. If it does not fit on one page, the process needs design work before it needs software. Build the automation in your own systems and document it, so your team can change a rule without calling whoever built it.
The savings come from removing handoffs. One client cut 93 hours of manual work a month from its Salesforce processes. Another saved 100+ hours on territory planning, and a third went from three systems to one native quote-to-revenue engine. Research automates too: scripted data extraction let one team identify 10,000 high-fit accounts its reps had been finding by hand.
Where AI fits
AI takes the work rules cannot: summarizing calls and CRM history, drafting follow-ups and answering pipeline questions in plain language. One client now builds its mutual action plans and proposals from its own Salesforce and Gong data. Start with one use case on one team, name an owner, and check the output against answers you know are right before you expand it.
Lean Layer does this work as a fractional RevOps, BI and AI enablement team that embeds in yours, giving you an entire RevOps org at the cost of a single manager. We are the #1 Revenue Operations Agency on G2 and trusted by 100+ revenue teams.
Frequently asked questions
What should I automate first in sales operations?
Start with recurring tasks that follow clear rules and take the most hours: reporting, lead routing, data cleanup and territory assignment. Fix the field definitions underneath them first, because automating on top of inconsistent data only spreads the errors faster.
Can I use AI to reduce manual sales work right away?
You can, but AI treats whatever is in the CRM as fact, so duplicates and stale fields turn into wrong answers. Clean the data and stabilize the core workflows first, then add AI one use case at a time and check its output against answers you know are right.
How do I measure whether automation is working?
Record a baseline before changing anything: hours per task, error rates and how long reports take to produce. Compare against it after each change. Lean Layer clients have documented savings of over 40 hours a month on reporting and 93 hours a month on Salesforce processes.
Do I need to hire a full-time RevOps person to do this?
Not necessarily. A fractional team brings data, workflow, tooling and enablement skills for the cost of one manager. A full-time hire makes sense when the workload is steady and narrow enough for one person to cover.