Step by step guide to setting up lead scoring in Einstein CoPilot for B2B sales teams

Sales teams get buried by leads that’ll never buy. If you’re in B2B and using Salesforce, you’ve probably heard about “AI-driven lead scoring” more times than you care to count. But setting it up—so it’s actually useful and not just more noise—is another story. This guide is for hands-on sales ops folks and admins who want lead scoring in Salesforce to work (not just tick a box for management). We’ll walk through setting up lead scoring in Einstein CoPilot, what’s worth your time, and what to skip.

Step 1: Get Your Data House in Order

Einstein CoPilot is only as smart as the data you feed it. If your leads are riddled with missing fields, old info, or junk entries, no AI is going to fix that. Before you touch any buttons:

  • Check required fields: At minimum, make sure fields like Email, Company, Industry, Lead Source, and Status are populated for most leads.
  • Clean up duplicates: Use Salesforce’s built-in duplicate management or a third-party tool.
  • Purge obvious junk: Leads with emails like asdf@asdf.com aren’t helping anyone.
  • Pro tip: If you don’t trust your data, don’t bother with lead scoring yet. Garbage in, garbage out.

Step 2: Enable Einstein Lead Scoring

Let’s cut through the fluff. Einstein CoPilot isn’t magic—it’s Salesforce’s AI layer that crunches your CRM data and tries to predict which leads are worth your team’s time.

Here’s how to turn it on:

  1. Check your edition and licenses: You need Salesforce Enterprise or higher, and Einstein features may be an extra paid add-on. If you don’t see Einstein in Setup, talk to your admin or AE.
  2. Go to Setup in Salesforce. Search for “Einstein Lead Scoring.”
  3. Click “Get Started” and follow the prompts.
  4. Select which lead record types to score. Most B2B teams just have one, but make sure.
  5. Set field permissions: Make sure users who need to see scores can view the fields.

Worth knowing: Einstein Lead Scoring uses your historical lead conversion data to train its model. If you don’t have at least a few hundred converted leads, the predictions will be weak. Salesforce says 1,000+ leads is best. If you’re a startup with 40 closed leads, you might want to hold off—or at least take the scores with a grain of salt.

Step 3: Tell Einstein What Matters (and What Doesn’t)

Einstein will try to pick up patterns in your data, but it’s not a mind reader. You can (and should) tell it which fields to ignore. For example:

  • Ignore fields that change post-conversion: Stuff like “Last Activity Date” or “Converted Date” will skew things.
  • Exclude marketing-only fields: If your “Lead Score” from another system is just marketing fluff, leave it out.
  • Custom fields: If you have custom fields that only get filled for a subset of leads, decide if they’re trustworthy.

To do this:

  1. In Setup, under Einstein Lead Scoring, go to “Field Exclusions.”
  2. Add any fields you don’t want Einstein analyzing.

Pro tip: Less is more. The more noise you feed the model, the less useful the score.

Step 4: Let Einstein Crunch the Numbers

Once enabled, Einstein will start chewing through your data. This can take anywhere from a few hours to a day or two depending on your org size.

  • Don’t expect instant magic. The first score is often “meh.” Give it a week or two to catch up with new data and retrain.
  • You’ll see a new “Lead Score” field added to your Lead record layout (if not, you can add it manually).

Step 5: Add Scores to List Views and Reports

Scores aren’t helpful if your sales team can’t see or use them. Make the “Lead Score” visible where it matters:

  • Add it to Lead List Views: So reps can sort or filter by score.
  • Add to page layouts: Front and center on the Lead page.
  • Build a report: Show leads by score bucket (e.g., 80+ is “hot”).

Pro tip: Don’t hide the model’s “Top Positive/Negative Factors” fields. These show why Einstein thinks a lead is good or bad—great for coaching reps and debugging the model.

Step 6: Create Simple Rules (Don’t Get Fancy Yet)

It’s tempting to build complex automation off your new lead scores. Don’t. At least, not at first. Keep it simple:

  • Auto-assign hot leads: Route leads with scores over a certain threshold to your best reps or a priority queue.
  • Flag dead leads: Leads with very low scores can go to nurture or be closed out.
  • Set up basic alerts: Let reps know when a new “hot” lead comes in.

Skip the 12-step flows until you’ve seen how the scores behave for a month or so. You’ll spot patterns and edge cases that no AI can predict.

Step 7: Train (and Listen to) Your Sales Team

AI is supposed to help humans, not replace them. Most reps are skeptical of new scores—sometimes for good reason. Here’s what works:

  • Explain the score: Show your team what factors drive high or low scores.
  • Encourage feedback: If reps spot obviously wrong scores, have them flag examples. This helps you spot data or setup issues.
  • Don’t force adoption: Let the best reps use the score as a supplement, not a replacement for their spidey sense.

Step 8: Review and Tweak Regularly

No lead scoring model is “set it and forget it.” Every few weeks:

  • Check score accuracy: Are high-scoring leads actually converting? Are low scores being ignored? Build a simple report to check.
  • Update field exclusions: If you add new fields or processes, adjust what Einstein analyzes.
  • Retrain as needed: If your business changes (new product, new segment), retrain the model.

Pro tip: Don’t chase perfection. A useful-but-flawed score beats no score at all.

What Works, What Doesn’t, and What to Ignore

What works: - Using lead score as a starting point for rep conversations—not as gospel. - Regularly cleaning your lead data. AI hates dirty data. - Reviewing the “why” behind scores, not just the number.

What doesn’t: - Blindly automating off the score. You’ll end up with good leads missed and junk getting through. - Treating the model as a black box. If you don’t know what fields drive scores, you’re flying blind. - Ignoring user feedback. Your reps will spot dumb patterns faster than the AI.

Ignore: - Fancy dashboards until basics work. - Vendor hype about “AI transforming sales overnight.” It’s just a tool—not a miracle.

Keep It Simple, Iterate, and Don’t Sweat Perfection

Lead scoring in Einstein CoPilot can help your B2B sales team focus on the right leads—but only if you keep it simple and stay skeptical. Start with the basics, check your work, and tune as you go. The best lead scoring setups are never “done”—they’re just quietly making your team’s life a bit easier, one lead at a time.