Salesforce AI lead scoring means Einstein Lead Scoring, a model that studies the Leads you converted and scores each open Lead on how much it looks like them. The catch is the history it needs. Salesforce asks for at least 1,000 Leads created in the last 200 days, 120 of them converted. Below that, if you score all Leads together, Einstein falls back to a global model built from many Salesforce orgs' anonymized data, and the score then describes somebody else's funnel.
This guide covers both routes: what Einstein needs and how to turn it on, then an AI fit score that judges every Lead against the customer you describe and writes the reason into a Salesforce field. The fit score needs no conversion history, so it works from the first Lead. If the scoring models themselves are new to you, start with B2B lead scoring models and examples.
To try the second route on your own org as you read, connect Salesforce to a free workspace and import a hundred Leads. A hundred is the Free plan's limit per table, and Free comes with 200 credits a month.
How Salesforce AI lead scoring works
Einstein Lead Scoring is predictive analytics pointed at your Lead object. It compares Leads that converted with Leads that did not and finds the field values that separate them. Each open Lead then gets a score for how closely it resembles the converted ones.
That is the predictive lead scoring Salesforce sells, and its strength is real: nobody has to guess the weights. A points model asks you to decide that a director title is worth 10 points. Einstein works that out from what actually happened in your org.
The same design sets its limits:
It learns from conversions, so it needs plenty of them. A young org, a new market, or a new product line has too few for the model to learn from.
It reads fields on the Lead. What the company sells or how many people it employs reaches the model only if somebody typed it into a Lead field.
It predicts conversion, not fit. If last year's conversions came mostly from one channel, the model probably learns the channel more than the customer.
Two neighbors share the name. Einstein Opportunity Scoring ranks open opportunities, not Leads. And Account Engagement has its own Einstein scoring for prospects, inside the marketing product formerly called Pardot, separate from the Sales Cloud feature this guide covers.
A complaint while we are on names. "AI lead scoring" now gets printed on almost any score, points tables with a new label included. The useful question to ask of any score, Salesforce's included, is what the model learned from and how many examples it had.
Salesforce AI lead scoring requirements
Here is what each route asks of your org before the first score appears:
Requirement | Einstein Lead Scoring | AI fit score in a custom field |
|---|---|---|
Salesforce edition | Sales Cloud Einstein: in Performance and Unlimited, extra cost on Enterprise | Any edition with API access and custom fields on Lead |
History needed | 1,000 Leads from the last 200 days, 120 of them converted | None: it reads the ideal customer profile you write |
Below that | With all Leads scored together, a global model from many Salesforce orgs' anonymized data | The same scorer; a Lead with too little data gets no score and a list of what was missing |
What it reads | Lead fields: all of them by default, or the ones an admin picks | The company and title on the Lead, plus company facts from data sources |
First score | Up to 48 hours after you turn it on | When the run finishes |
Updates | Rebuilds the model about every 10 days | Each run, or as Leads arrive and change |
The Einstein column comes from Salesforce Help. Salesforce renames and repackages its editions often, so check its pricing page for the current packaging before you budget for the add-on.
How to enable Einstein lead scoring in Salesforce
If your org clears the bar, the setup is short, and an admin does it in Setup:
Type Einstein Lead Scoring in Quick Find and open it.
Turn it on, then choose default or custom settings. The default scores every Lead together and reads all Lead fields.
With custom settings, pick the conversion milestone, decide whether segments of Leads get their own model, and choose the fields to include.
Click Score Leads, then wait. The first analysis and scoring can take up to 48 hours.
Check that the Einstein Lead Scoring component sits on your Lead pages, then add the Lead Score field to the list views reps work from.
Labels shift between Salesforce releases, so treat these as landmarks more than exact button names.
One detail trips people up. If you picked the fields by hand, or split Leads into segments, a custom field you add later stays invisible to Einstein until an admin adds it to the included fields of each segment. Enriching Leads with new company fields does nothing for the Einstein score until that happens.
When the history is too thin: score against your ICP
Say your team started outbound in the spring. You have 400 Leads, most of them created by reps from lists, and 30 have converted. Einstein's bar is 1,000 and 120, so if you score them all together, every score on those Leads comes from the global model.
You could wait a year for the history to build up. Or you could ask a different question. Instead of which Leads resemble past conversions, ask which Leads match the customer you sell to. That question needs no history at all. It needs a written description of your ideal customer and enough facts about each company to hold up against it.
The facts are usually there to be found:
"Sales comes down to timing and to having the right data, and most of that data is already online."
Utku Zihnioglu, CEO of Oneprofile
In Oneprofile that question is the job of the ICP scorer. It reads each Lead against the Ideal customer profile in your business context and returns three things:
ICP fit: strong at 70 or above, possible from 40 to 69, poor under 40.
Fit score: a number from 0 to 100, calibrated to the band.
Fit reasons: the two or three facts that drove the decision, each naming the field it came from.
Here is a worked example. The ideal customer profile reads "B2B software companies with 20 to 200 employees that sell to finance teams; the buyer runs finance or operations". Three Leads come back:
Lead | What enrichment found | ICP scorer result |
|---|---|---|
Head of finance operations at cleardue.example | Invoicing software, 60 employees, sells to finance teams | strong, 84: B2B software, inside the employee range, and the title is the buyer |
Founder at granitecreative.example | Brand design agency, 8 employees | poor, 18: a services firm, and below the employee range |
Operations manager at otterstrategy.example | Nothing beyond the name: no description, industry, or employee count | No band and no score; Missing lists all three |
Look at the third Lead. With too little to go on, the scorer lists what it lacked instead of guessing, so a company nobody could check never lands in poor. A points model makes exactly that mistake with an empty field.
None of this says turn Einstein off. Who behaves like past buyers is still worth knowing. Whether the company was worth chasing at all is a different question, and the fit score answers it in words a rep can read.
Set up Salesforce AI lead scoring with a fit score and reason
Six steps, about 30 minutes, most of it waiting on the first run.
Create the score and reason fields on the lead
In Salesforce, open Object Manager, then Lead, then Fields & Relationships, and add three custom fields:
Salesforce field on Lead | Field type | What lands in it |
|---|---|---|
AI Fit Score | Number, 0 decimal places | A number from 0 to 100 |
AI Fit Band | Text | strong, possible, or poor |
AI Fit Reasons | Text Area (Long) | The two or three facts behind the score |
The export writes only to fields Salesforce already has and never creates one. If you add a field later, Refresh in the export editor fetches the field list again. Keep these fields separate from any Einstein or formula score, so nobody mistakes one number for another.
Connect Salesforce and import your leads
Open Connections in Settings, click New, find Salesforce, and click Connect Salesforce. Sign in with a user who can read and edit Leads. The connection reads and updates Leads through Salesforce's published APIs, so your org needs API access. The Salesforce REST API limits decide how many records each of those calls can read or write.
On Tables, choose Import from a tool. Pick Salesforce as the Tool and Lead as the Object. Match rows on picks the Salesforce Id for you; keep it, because it never changes. Under Output, tick Company, Website, Email, and Title. To try it on a small slice first, pick First and enter 100.
The ICP scorer needs a company domain, because a score attached to the wrong company of the same name is worse than none. Website is the usual source. Where reps left it blank, click Tools, open Functions, choose Write a formula, and describe what you want, such as the domain from the email address. A personal address like gmail.com gives no company domain, so those Leads wait for a website.
Fill company facts from the domain
Add Enrich company from domain and point it at the domain. It returns the company name, industry, employee count, headquarters city and country, founded year, and a description. It checks data sources one at a time, most accurate first, and stops at the first answer. It costs up to 15 credits a Lead, and a company lookup that finds nothing costs nothing. The data enrichment page explains the order.
This is the step Einstein never gets on its own: what the company sells and how big it is, on Leads where a rep typed only a name.
Score every lead against your ICP
Write your ideal customer under Ideal customer profile in Business context, one fact per bullet: segment, headcount range, and who they sell to. The roles you sell to go under Ideal buyer personas, which the scorer reads too. To work it out from your closed-won accounts, see how to score accounts against your ICP.
Then click Tools, open the Agent tab, and add the ICP scorer from Built-in agents. Point its inputs at the domain, company name, description, industry and employee count, and the Lead's Title as Job title. It costs 3 credits a Lead. Run the first 10 and read the reasons before you run the rest.
Export the fit score back to Salesforce
Click Export above the grid and pick Salesforce under Send to CRM. The setup opens on Create or update records. Click Change, pick Salesforce again, and choose Update records, so the export can only change Leads that already exist. Set the Object to Lead and Find the record by the Id.
Map Fit score to AI Fit Score, ICP fit to AI Fit Band and Fit reasons to AI Fit Reasons. Under When a record matches, keep Write mode on Change mapped fields only and turn on Skip empty values, so a rerun never blanks a score. Click Test on these rows to see what a real push would do without writing anything, then Create and push.
Refresh and score new leads automatically
Click Source at the bottom of the grid, turn on Schedule refreshes, and pick Daily. Each refresh updates the Leads already imported and brings in the ones created since.
Then open Table settings and, under Run settings, turn on Run automatically on new and changed rows, with Keep existing results on. New Leads get enriched and scored as they arrive, and a score that already succeeded is not paid for again. This is a standing spend, so add a run condition, such as only Leads with a domain, if your org takes in a lot of junk.
The export is the one step that waits for you. Open its run menu and choose Push all rows after each refresh. Skip unchanged rows, under Reliability, leaves out Leads whose score did not move.
Put the fit score to work in Salesforce
Anyway, a score in a field does nothing until something reads it. A few places to put it:
Page layouts and list views. Add AI Fit Score and AI Fit Band to the Lead page layout and to the list views reps work from, and sort by the score.
Routing. The export saves a new value to the field, so a record-triggered flow can watch AI Fit Score and assign Leads at 70 or above to a queue.
Beside Einstein. If you have both, show both. When they disagree on a Lead, read the reasons field first: it names the facts the fit score was built on.
Beside a formula. If reps want a points model they can audit, the formula fields in the step-by-step Salesforce lead scoring guide read the same enriched fields.
That is the version of Salesforce AI lead scoring worth showing a rep: a number, a band, and the facts behind it, on the Lead they already open every day.
On cost: scoring alone is 3 credits a Lead, so Free's 200 credits a month score 66 Leads that already carry their company facts. Free also allows 3 integrations, 3 tables, and 100 rows per table. Team is $20 a month for 1,000 credits and 10,000 rows per table, and extra credits are $20 per 1,000 on every paid plan. Score your first Salesforce Leads for free and see whether the reasons match what your reps already believe.

