Most scoring models lose sales’ trust within a quarter. A rep clicks into a “hot” lead, sees a student who downloaded a whitepaper for a class project, and closes the tab. Two weeks later, another MQL turns out to be a competitor’s marketing manager who registered for your webinar. After three or four of those, the rep stops opening the queue at all, and the label “MQL” becomes noise no matter how many point rules sit behind it.
That failure is rarely about the point math. It happens because the model rewards enthusiasm without checking fit, never subtracts anything, and never gets recalibrated against what actually closed.
You can build a scoring model and lifecycle in Adobe Marketo Engage that sales trusts, but only if you treat it as a living system: fit plus intent in the point model, negative scoring and decay running in the background, and a quarterly review against real closed-won data.
A quick naming note. The product is officially Adobe Marketo Engage, though practitioners still say “Marketo” in conversation. This piece uses both. Adobe now lists Marketo Engage among its Adobe CX Enterprise applications, the agentic AI platform it introduced at Adobe Summit in April 2026.
The Marketo Engage product name stays the same, and the features this article covers still work as described in Adobe’s documentation. The mechanisms referenced here (Smart Campaigns, Smart Lists, score fields, batch campaigns, program statuses for lifecycle tracking) are current in Adobe’s Marketo Engage documentation as of October 2026.
What follows covers building the point model, the lifecycle stages it feeds, and the operational discipline that keeps both credible six months after launch.
In this guide:
- How scoring and lifecycle work together
- Build a scoring model around fit and intent
- Configure the model in Adobe Marketo Engage
- Create a lifecycle process sales will follow
- Keep the model accurate as buyer behavior changes
- Frequently asked questions
- A trusted model turns engagement into better sales conversations
How scoring and lifecycle work together
Scoring answers “how interested and how qualified is this person,” while lifecycle answers “who owns this person right now and what happens next.” Confusing the two is why so many models produce leads sales won’t touch.
What lead scoring is designed to decide
A score ranks people, so a rep working 40 leads knows which five to call first. Adobe’s own guidance on building a person scoring model frames it as a way to measure lead qualification and set criteria for sales readiness.
The score itself does not assign work. It informs a decision.
Why a high score does not automatically make an MQL
A lead can hit 100 points by visiting your blog 30 times and opening every email you send. If that person works at a 4-person agency and your ICP starts at 200 employees, the score is measuring reading habits.
Good models require both a fit threshold and a behavior threshold before the lifecycle stage changes. Someone at a target account who requests a demo clears both. A high-enthusiasm, low-fit lead stays in nurture.
The roles of MQL, SAL, SQL, and recycled leads
Each stage marks a change in ownership or commitment:
- MQL: Marketing says this person meets the agreed bar. Marketing still owns the record until a rep touches it.
- SAL (Sales Accepted Lead): A rep has looked at the lead and agreed to work it. Ownership transfers here.
- SQL: The rep has had a real conversation and confirmed need, timing, or budget. This is where pipeline starts.
- Recycled: Sales worked the lead and found it too early or unresponsive. It goes back to marketing nurture with a reason code, not into a dead pile.
Without the SAL step, you cannot tell the difference between “sales rejected this lead” and “no one ever called.”
Build a scoring model around fit and intent
A working model keeps two scores separate: one for behavior (intent) and one for demographics and firmographics (fit). Adobe’s own current guidance on building a person scoring model recommends this same split, plus a Total Score that combines them.
The split is diagnostic. When sales complains about lead quality, you need to see whether the bad leads scored high on behavior with weak fit, or high on fit with almost no engagement. A single blended number hides that.
Behavioral signals that show buying intent
Behavior is what the person does, but not all behavior means the same thing.
High-intent actions:
- Requests a demo or contacts sales through a form
- Visits the pricing page (especially more than once in a week)
- Attends a webinar live, as opposed to registering and not showing
- Downloads a comparison guide, implementation checklist, or ROI calculator
- Opens and clicks three or more emails within seven days
- Visits a product documentation or integrations page
Low-intent actions that still count for something:
- Opens a newsletter
- Clicks a footer link
- Reads a top-of-funnel blog post
- Follows your company page
The gap between a demo request and a footer click is enormous. Your point values should reflect that gap honestly.
Demographic and firmographic criteria that define ICP fit
Fit is what the person and their company are, independent of anything they clicked.
Person-level criteria:
- Job title seniority: Director, VP, or C-level in the relevant function
- Department: Does this person sit in the function that uses or buys your product
- Country or region: Do you sell there, and can you support it
Company-level criteria:
- Employee count band: 200 to 1,000 employees, for example, if that is your sweet spot
- Industry: SaaS, financial services, healthcare, whatever your closed-won data supports
- Annual revenue band, where you have the data
- Tech stack signals, if you enrich for them (a company already running a CRM you integrate with)
Most of this comes from form fields plus data enrichment. If your form only asks for email and name, your fit score has nothing to work with, and that is a data problem before it is a scoring problem.
Behavioral scoring vs. lead grading
Lead grading is the fit half of the equation, often expressed as a letter (A through D) instead of a number. Scoring is the intent half.
| Behavioral scoring | Demographic scoring (lead grading) | |
| What it measures | Interest and buying intent | Fit against your ICP |
| Data source | Web activity, email engagement, form fills, event attendance | Form fields, CRM data, enrichment vendors |
| Example signals | Pricing page visit, demo request, webinar attendance, three email clicks in a week | VP-level title, 500-employee company, SaaS industry, US-based |
| How fast it changes | Constantly, sometimes hourly | Rarely, only when the person changes jobs or the company grows |
| Needs decay | Yes | No |
| Marketo mechanism | Trigger-based Smart Campaigns on Visits Web Page, Fills Out Form, Clicks Link in Email | Trigger and batch Smart Campaigns on Data Value Changes, plus filters on title and company fields |
| Failure mode if used alone | High-scoring students, competitors, and job seekers | Perfect-fit accounts who have never heard of you |
The classic breakdown: someone visits your site 22 times, opens 14 emails, and hits 140 points. A rep calls and finds a freelance consultant researching a blog post. The model measured enthusiasm and never asked whether this person could sign a contract.
An illustrative starting-point scoring matrix
These numbers are a starting point, not a formula to copy. Every model needs recalibration against your own conversion data within the first two quarters.
| Signal | Type | Illustrative points |
| Requests a demo or free trial | Behavior | +30 |
| Visits pricing page (per visit, max 3) | Behavior | +12 |
| Attends webinar live | Behavior | +15 |
| Registers for webinar, no-show | Behavior | +3 |
| Downloads mid-funnel asset (comparison guide, ROI tool) | Behavior | +10 |
| Downloads top-funnel asset (blog, ebook) | Behavior | +3 |
| Clicks link in email | Behavior | +2 |
| Opens email | Behavior | +1 |
| Visits careers page | Behavior | -10 |
| Title contains VP, Director, Head of, or C-level | Fit | +20 |
| Title in target department | Fit | +15 |
| Company size 200 to 1,000 employees | Fit | +20 |
| Company size under 25 employees | Fit | -15 |
| Industry matches ICP | Fit | +15 |
| Country outside your sales territory | Fit | -20 |
Set your MQL threshold before you build any campaigns. A common structure requires a minimum on both scores, for example, a Behavior Score of 50 or higher and a Demographic Score of 40 or higher, so neither dimension can carry a lead across the line alone.
Calibrate after you have data. Pull your last 12 months of closed-won deals and export each contact’s behavior history before the opportunity was created. Look for which actions show up repeatedly across won deals and rarely across lost ones. If pricing page visits appear in 78% of closed-won journeys and only 12% of closed-lost, that behavior deserves more weight than your original guess.
An AI assistant helps here in a concrete way: export closed-won and closed-lost contact activity to CSV, then prompt the assistant to compare frequency of each activity type across the two groups and flag the largest gaps.
You still make the point-value decisions, but you start from a ranked list instead of intuition. Once your manual model has enough history, Marketo’s Predictive Audiences feature can extend the same idea automatically.
This breakdown of Marketo AI and predictive scoring covers the data volume it actually needs (a practitioner baseline of roughly 400 historical conversions) before it’s worth turning on.
Configure the model in Adobe Marketo Engage
Build the scoring model as an operational program with separate campaigns for each scoring dimension, then use tokens so you can adjust point values in one place. Adobe’s lead scoring program tutorial walks through this structure, including how to import pre-built campaigns.
Score fields and data governance
Create three custom score fields: Behavior Score, Demographic Score, and a Total Score that sums them. Marketo Engage ships with a default Person Score field, and many teams keep it as the total while building custom fields for the two components.
Store point values as Score-type My Tokens at the program or folder level, for example, a token named {{my.Pricing Page Visit}} with a value of +12. Score is a dedicated My Token type that Adobe built for the Change Score flow step (see Adobe’s My Tokens documentation), so each scoring campaign references the token instead of a hard-coded number. When you decide a pricing page visit is worth 15 instead of 12, you edit one token instead of hunting through campaigns.
Set governance rules early:
- Only the scoring program changes score fields. No other campaign touches them.
- Keep every scoring Smart Campaign in one operational program, not scattered across marketing programs.
- Document each point value and its rationale in the campaign description field so the next person inherits context.
Smart Campaigns for meaningful engagement triggers
Trigger-based Smart Campaigns handle behavior scoring in real time. Each one uses a trigger, optional filters, and a Change Score flow step.
Examples:
- Trigger: Visits Web Page is /pricing. Flow: Change Score, Behavior Score, +12. Add a Smart Campaign setting limiting each person to 3 qualifications so a single obsessive visitor cannot run the score up.
- Trigger: Fills Out Form is “Demo Request.” Flow: Change Score, +30, then Change Program Status.
- Trigger: Program Status is Changed, new status “Attended.” Flow: Change Score, +15. A separate campaign scores “Registered” at +3.
- Trigger: Clicks Link in Email. Flow: Change Score, Behavior Score, +2, matching the scoring matrix above.
Demographic scoring uses a Data Value Changes trigger on Job Title, Number of Employees, or Industry, plus a batch campaign that runs against your existing database when you add a new criterion.
Smart Lists for qualification and routing logic
Smart Lists define who qualifies without changing anything. Build one for each threshold you care about:
- MQL Ready: Behavior Score ≥ 50 AND Demographic Score ≥ 40 AND Lifecycle Stage is Inquiry
- High Fit, Low Engagement: Demographic Score ≥ 40 AND Behavior Score < 20 (feed this to nurture or an outbound list)
- High Engagement, Low Fit: Behavior Score ≥ 50 AND Demographic Score < 20 (review this list monthly; it tells you whether your fit criteria are too narrow)
The last list is your early warning system. If it fills up with real prospects, you need to widen your ICP definition.
Batch campaigns for score decay and data cleanup
Batch campaigns run on a schedule against a Smart List, which makes them the right tool for anything time-based:
- Run a nightly or weekly batch that finds people whose Behavior Score hasn’t changed in the last 30 days (the Not Score Was Changed filter, covered in the decay fix below) and subtracts points
- Run a monthly batch that clears stale scores on records inactive for a year
- Run a batch that flags records missing critical fit fields, so you know where enrichment is failing
Create a lifecycle process sales will follow
Lifecycle stages only work when each one has entry criteria a person can check in five seconds and a single owner. Adobe’s guidance on building a person lifecycle program uses program statuses in an operational program to track stage progression, with Smart Campaigns moving people between statuses.
Copying a generic funnel is the most common mistake here. If your sales team runs an SDR-to-AE handoff, your lifecycle needs a stage for it. If they do not, forcing one in creates a stage nobody updates.
Define stage entry criteria and ownership
| Stage | Definition | Entry criteria | Owner |
| Inquiry | Known person in the database | Email address captured through any source | Marketing |
| MQL | Meets the agreed scoring bar | Behavior Score ≥ 50 AND Demographic Score ≥ 40, or an instant-qualify action like a demo request | Marketing |
| SAL | A rep has reviewed and accepted the lead | Rep sets Lead Status to Accepted within the SLA window | Sales (SDR) |
| SQL | Rep has confirmed need, timing, or authority in a live conversation | Rep logs a completed discovery call and sets Status to Qualified | Sales (SDR or AE) |
| Opportunity | Deal created in CRM with a value and close date | Opportunity record created and linked to the contact | Sales (AE) |
| Customer | Closed-won | Opportunity stage set to Closed Won | Sales (AE) |
| Recycled | Worked by sales, sent back to nurture | Rep sets Status to Recycled with a reason code | Marketing |
| Disqualified | Never a fit | Rep sets Status to Disqualified with a reason code | Sales |
Rename these to match what your reps say out loud. If your team calls SQL a “qualified opp,” use that. Update stage names people already use, as invented names get ignored.
Set an SLA for follow-up, acceptance, and rejection
The SLA is the agreement that makes MQLs mean something. Write it down and get both leaders to sign it.
A workable structure:
- Marketing commits to a monthly MQL volume and a minimum fit standard
- Sales commits to first touch within a defined window, for example, 24 business hours for standard MQLs and 1 hour for demo requests
- Sales commits to setting Lead Status to Accepted, Recycled, or Disqualified within 5 business days
Enforce it with a Smart Campaign that triggers when Lead Status hasn’t changed within the window and alerts the rep’s manager. Track SLA compliance as a reported number, not a hope.
Capture rejection reasons and lead recycling rules
Create a required “Disqualification Reason” picklist in your CRM with options that are useful for scoring: Wrong Title, Company Too Small, Competitor, No Budget, Bad Timing, No Response, Already a Customer.
Route each reason differently:
- Send Bad Timing or No Budget to Recycled, reset the behavior score to a lower baseline, and enroll them in a long-cycle nurture. A Smart Campaign re-scores them when they return.
- No Response after a defined number of attempts goes to Recycled with a 60-day hold before the lead can re-MQL.
- Wrong Title or Company Too Small go to Disqualified and feed straight into your fit-score review. If Wrong Title shows up 40 times a quarter, your title scoring is wrong.
- Competitor goes to Disqualified permanently, with a Smart Campaign adding the domain to a suppression list.
Review the reason distribution monthly. It is the most direct feedback you will ever get on the model.
Prevent duplicate routing and stale handoffs
Add a filter to your MQL campaign excluding anyone whose Lifecycle Stage is already MQL, SAL, SQL, Opportunity, or Customer. Without it, a customer who downloads a guide gets routed to an SDR who wastes an hour.
Set a re-MQL waiting period. If a lead was recycled 20 days ago, they should not re-MQL on a single email click. A Smart Campaign filter on “Lifecycle Stage was changed to Recycled in past 60 days” blocks it.
Run a weekly batch that finds records sitting in SAL for more than 14 days with no logged activity and pushes them back to Recycled. Stale records in a rep’s queue are invisible pipeline rot.
This kind of routing logic gets much harder once Marketo syncs bidirectionally with Salesforce; this guide to Marketo-Salesforce sync architecture and dedup covers the failure modes that corrupt lifecycle stage data at scale.
Keep the model accurate as buyer behavior changes
Recalibrate on a fixed schedule, quarterly for most teams, using conversion data rather than opinions. A model built in January against 2025 buying behavior does not describe how people buy by September.
Measure conversion rates by score band and lifecycle stage
Group leads into score bands and measure what happened to each band:
Quarterly score band worksheet. Run this once a quarter for each of four bands: 90+, 70 to 89, 50 to 69, and below 50.
- Pull every lead that entered the band during the quarter, using the Total Score at the moment it reached the band.
- Record the MQL to SAL rate: the share of the band’s MQLs that a rep accepted.
- Record the SAL to SQL rate: the share of accepted leads that reached a qualifying conversation.
- Record the SQL to closed-won rate: the share of SQLs that became customers.
- Note the change from last quarter for each rate, so you can see drift before sales feels it.
- Compare the bands against each other using the two checks below.
If the 50 to 69 band converts to SQL at nearly the same rate as the 90+ band, your threshold is set too high, and you are sitting on leads sales would happily work. If the 90+ band converts no better than the middle, some high-value behavior in your model is not predictive.
Also track stage-to-stage velocity. A lead that sits in MQL for 11 days on average points at an SLA problem, not a scoring problem.
Review sales feedback alongside revenue outcomes
Hold a 30-minute monthly call with two or three reps who work your leads. Bring the disqualification reason report and five specific lead records: two that closed, three that got rejected.
Ask what they saw on the rejected records that the model missed. Reps notice patterns before your reports do, like a spike in leads from a certain job function or a webinar that keeps producing tire-kickers.
Pair that with the numbers. Anecdote alone leads to overcorrecting after one bad lead; data alone misses signals that are not in your fields yet.
Recalibrate thresholds, point values, and decay rules
Change one thing at a time and document it. Common adjustments are:
- Threshold: Raise or lower the MQL bar based on conversion by band and on sales capacity. If reps can work 100 leads a month and you are sending 300, the bar is too low.
- Point values: Increase weight on behaviors that appear in closed-won journeys and cut behaviors that appear equally in won and lost.
- Decay rate: If leads regularly re-engage at day 45, a 30-day decay is too aggressive. Adjust the batch campaign’s inactivity filter.
- Score caps: Cap repeatable behaviors so no one reaches MQL through volume alone.
Keep a simple changelog with the date, what changed, and why. When someone asks in six months why a demo request is worth 30 points, the answer should exist in writing.
Common scoring failures and how to correct them
Scoring every micro-interaction the same
A footer click worth the same as a demo request produces MQLs made of newsletter readers. Fix: rebuild point values in tiers, with high-intent actions worth 5 to 10 times a passive one.
No negative scoring
Nothing ever comes off the record, so scores only climb. Fix: build Smart Campaigns that subtract for a careers page visit, an unsubscribe (set Behavior Score to a floor value and change Lead Status), a free email domain on a B2B model, and a title outside the buying committee.
Add a Smart Campaign that matches known competitor domains and immediately sets Lifecycle Stage to Disqualified.
No decay
A lead who scored 120 in March and has done nothing since still shows as hot in September. Fix: a weekly batch campaign with a filter on “Not Score Was Changed: Behavior Score, in past 30 days” that subtracts 10 points, running until the score reaches a floor. This filter is cleaner than “Not Activity: Any” because it only looks at activity your model actually scores.
“Any activity” also counts things the lead never did, like an email being sent to them or a field updating during a CRM sync, so leads who have gone quiet keep looking active and never decay. Because the decay step is itself a score change, each lead loses points at most once per 30-day window, which gives you a steady decline instead of a sudden drop. Skip this and your “hot leads” list ends up half stale, which is precisely how reps learn to ignore it.
Never recalibrating
The model launches, gets celebrated, and never gets touched. Fix: put a recurring quarterly review on the calendar with a named owner. If no one in-house has the Marketo depth to own it, this comparison of the best platforms to hire Marketo experts and consultants covers where to find a specialist who can.
Vague stage criteria
“Sales Qualified” defined as “the rep thinks it’s good” means every rep applies a different bar and your funnel reporting is fiction. Fix: write entry criteria that a person can verify by looking at a record.
Scoring on data you do not have
A fit model built on company size when 70% of records have a blank employee count scores almost nobody. Fix: audit field completeness first, then either add form fields, add enrichment, or build the model on data you reliably capture.
Frequently asked questions
What is a good lead score threshold for MQL in Marketo?
There’s no universal number, and any article that gives you one flat threshold is guessing. A workable starting structure requires a minimum on both dimensions, for example, a Behavior Score of 50 or higher and a Demographic Score of 40 or higher, so neither enthusiasm nor fit alone can push a lead across the line.
Treat that as a hypothesis to test, not a target: pull your last 12 months of closed-won deals, see what score bands they actually came from, and adjust your threshold to match what your own conversion data shows.
What is the difference between MQL, SAL, and SQL?
An MQL means marketing believes the person meets the agreed bar, but marketing still owns the record. An SAL (Sales Accepted Lead) means a rep has reviewed the lead and agreed to work it, which is where ownership transfers. An SQL means the rep has had a real conversation and confirmed need, timing, or budget, which is typically where pipeline starts. Skipping the SAL step is a common mistake, because without it you can’t tell the difference between “sales rejected this lead” and “no one ever called.”
How often should you recalibrate a Marketo lead scoring model?
Quarterly for most teams. A model built against last year’s buying behavior doesn’t necessarily describe how people buy today, and scoring models that never get touched after launch are one of the most common reasons sales stops trusting them.
Recalibration means checking conversion rates by score band, reviewing disqualification reasons with sales, and adjusting point values, thresholds, or decay rates based on what the data actually shows, not guessing at a schedule.
What is the difference between lead scoring and lead grading?
Lead scoring measures behavior and intent (what someone does: page visits, email clicks, demo requests) and changes constantly. Lead grading measures fit against your ideal customer profile (job title, company size, industry) and is often expressed as a letter rather than a number, since it rarely changes unless someone switches jobs or their company grows. Blending the two into one number without separating them makes it much harder to diagnose why lead quality is bad.
Why does sales stop trusting MQLs?
Usually because the scoring model rewards enthusiasm without checking fit, and because nothing ever comes off a lead’s score once it’s added. A student who downloads a whitepaper for a class project, or a competitor’s employee who registers for a webinar, can rack up a high score through sheer activity with no buying intent behind it.
After a rep opens a few “hot” leads that turn out to be unqualified, they stop opening the queue altogether, and the MQL label becomes noise no matter how sophisticated the underlying point system is.
What is score decay and why does it matter?
Score decay is the practice of reducing a lead’s score over time when they’ve gone inactive, typically through a scheduled batch campaign that checks for no activity within a set window (30 days is a common starting point) and subtracts points accordingly.
Without it, a lead who scored high six months ago but has gone completely silent still shows up as “hot” today. That’s exactly how a “hot leads” list ends up half stale, which is one of the fastest ways to teach a sales team to stop trusting it.
A trusted model turns engagement into better sales conversations
A scoring model earns trust through the parts nobody demos: negative scoring that keeps competitors and job seekers out of the queue, a batch campaign quietly decaying stale scores every week, lifecycle stages with criteria a rep can check in seconds, and a quarterly recalibration against what actually closed.
Separate fit from intent so you can diagnose problems. Require both thresholds before an MQL fires. Write an SLA both teams sign, capture disqualification reasons, and feed those reasons back into the point values. Frame every number as a starting point, because the first version of your model is a hypothesis about how people buy.
Sales will tell you when the model is wrong. The job is building the operational habits that let you hear it and respond within a quarter, not a year.
A scoring model that sales trusts takes a Marketo specialist who treats it as a living system tied to real conversion data. Arc pre-vets Marketing Operations specialists for technical depth and English fluency before you see a profile, and HireAI matches your requirements against a pool of 450,000+ vetted professionals across 190 countries to return a shortlist in minutes.
