How is outbound lead scoring different from inbound scoring?
Inbound lead scoring grew up in marketing automation. It adds points when someone downloads a guide, opens emails or visits the pricing page, then hands the lead to sales above a threshold. It measures engagement with you.
Outbound prospects have not engaged with you yet, so engagement points are mostly zero. An outbound score has to rely on what you can observe from outside: how well the company matches your ideal customer profile, what public evidence of need exists, and how recent it is. It answers "should we reach out, and how soon?" rather than "is this lead ready for sales?"
Scoring is also different from qualification. Frameworks such as BANT (budget, authority, need, timing) are applied during a conversation. A score is the bet you place before any conversation exists.
What types of lead scoring are there?
"Lead scoring" covers several methods that measure different things. Knowing which one you are building avoids the most common confusion: giving engagement points to people who have never heard of you.
| Type | What it measures | Data it needs | Best for | Main limit |
|---|---|---|---|---|
| Fit (explicit) | How closely a company and contact match your profile | Industry, size, role, region, tech stack | Every motion; the base layer of any model | Says nothing about timing |
| Engagement (implicit or behavioral) | What a lead does with your emails, site and content | Marketing automation or website tracking | Inbound and nurture programs | Close to zero for cold prospects |
| Signal or intent | Evidence of a need from outside your own channels | Public posts, company events, intent data | Outbound and account prioritization | Quality varies and signals age fast |
| Predictive | Similarity to past leads that converted, learned by a model | Hundreds of historical outcomes | Teams with enough closed deals to learn from; platforms such as 6sense apply it to accounts | Hard to explain, weak with little history |
| Account | The combined picture across everyone at one company | Contact data rolled up by company | Selling to a buying committee | Needs clean matching of people to companies |
Score or grade?
Some platforms keep fit and engagement apart. Salesforce Account Engagement (formerly Pardot), for example, shows a numeric score for engagement and a letter grade for how well the prospect matches your profile. An "A" grade with a low score is a good company that has not engaged yet, which is exactly the group outbound exists to reach.
MQL, SQL and where outbound fits
A marketing-qualified lead (MQL) has crossed an engagement threshold. A sales-qualified lead (SQL) has been accepted by sales after a first check, usually a conversation. Outbound leads skip the MQL stage: the score decides whether to reach out, and the first reply or call decides whether the lead becomes sales-qualified.
What are fit, intent and timing?
Fit: could they buy?
- Industry, business model and company size within your target range.
- A decision-maker you can identify, with a verified business email.
- Region, language and anything that rules a company out (current customer, open deal, competitor, partner).
Intent: is there evidence of need?
- Explicit statements: a request for a tool, a search for an alternative, a complaint about a competitor.
- Events that create need: a launch, a funding round, a hire for a related role, a tech change. See 15 buying signals with examples.
- First-party behavior if you have it, such as visits to your pricing page or a past conversation.
Timing: why now?
- The age of the signal. A request posted yesterday and the same request from two months ago are different leads.
- Signals that stack: two independent events within a few weeks point to a real change.
- Known windows, such as the first months of a new leader or the weeks after a funding announcement.
A sample outbound lead scoring model
Use this as a starting point and adjust the points to your market. Fit is capped at 40 points, intent at 35 (take the strongest signal, plus a bonus for a second one) and timing at 25. Exclusions override everything.
| Criterion | Part | Points |
|---|---|---|
| Industry in your target list | Fit | +10 |
| Company size in your target band | Fit | +10 |
| Decision-maker identified, business email verified | Fit | +10 |
| Sells to businesses (B2B model) | Fit | +5 |
| Region and language you serve | Fit | +5 |
| Asked publicly for a tool like yours | Intent | +30 |
| Looking for an alternative to a named competitor | Intent | +30 |
| Complained about a competitor | Intent | +25 |
| Described the pain point you solve | Intent | +20 |
| Company event tied to your offer (launch, funding, related hire, tech change) | Intent | +15 |
| General interest in your category only | Intent | +5 |
| Second independent signal in the last 30 days | Intent | +5 |
| Signal less than 48 hours old | Timing | +25 |
| Signal 3 to 7 days old | Timing | +15 |
| Signal 8 to 30 days old | Timing | +5 |
| Signal older than 30 days | Timing | 0 |
| Company size outside your band | Penalty | −10 |
| Consumer or personal use, not a business | Penalty | −30 |
| Current customer, open deal, competitor, partner or past unsubscribe | Exclusion | Do not contact |
What should each score trigger? A worked example
A score only helps if each band leads to a different action. Sample thresholds for the 100-point model:
- 70 and above: contact within 24 hours with a message built around the signal; review by hand if the account is large.
- 50 to 69: contact within the week with lighter personalization, or wait for a second signal.
- 30 to 49: keep on a watch list and re-score when a new signal appears.
- Below 30: do not contact. If many leads land here, check whether your profile or your source is producing noise.
| Lead | Fit | Intent | Timing | Score and action |
|---|---|---|---|---|
| A: 30-person B2B SaaS; the founder asked on Reddit yesterday for a cold email tool; email verified | 40 | 30 | 25 | 95: contact today and reference the request |
| B: 400-person software company outside your size band; Series B three weeks ago; head of sales verified | 20 (after the size penalty) | 15 | 5 | 40: watch list until a stronger signal appears |
| C: 12-person agency; complained about a competitor five days ago; owner found, email not yet verified | 30 | 25 | 15 | 70: verify the email first, then contact this week |
| D: same profile and signal as lead A, but the company is in an open deal with your sales team | n/a | n/a | n/a | Excluded: no score. Send the signal to the account owner, who can use it in the open deal. |
| E: solo founder asks publicly for a cheap email tool for a personal newsletter; no business email | −5 (region only, after the size penalty) | 30 | 25 | 20 after the −30 personal-use penalty: do not contact, however fresh the request |
Reach companies with a reason to buy this week
Startories finds the buying signal, verifies the decision-maker and runs the outreach until they book a call.
How do you build the model in a spreadsheet?
You do not need software to start. One sheet with a row per lead and a column per criterion runs the whole model, and the formulas below work in both Google Sheets and recent versions of Excel. The column letters are an example layout; adjust them to yours.
Set up the input columns
A: company. B: industry match (Y or N). C: size in band (Y or N). D: decision-maker verified (Y or N). E: B2B model (Y or N). F: region served (Y or N). G: signal type, picked from a dropdown. H: signal date. I: date of a second, independent signal (blank if none). J: personal-use flag (Y or N). K: exclusion flag (Y or N).
Keep the signal points on their own tab
On a tab named Points, list each signal type in column A and its points in column B: tool request 30, alternative search 30, competitor complaint 25, pain point 20, company event 15, category interest 5. Changing a weight then means editing one cell, not every formula.
Compute fit, intent and timing
Fit in L: =IF(B2="Y",10,0)+IF(C2="Y",10,-10)+IF(D2="Y",10,0)+IF(E2="Y",5,0)+IF(F2="Y",5,0). Intent in M: =IFERROR(VLOOKUP(G2,Points!A:B,2,FALSE),0)+IF(AND(I2<>"",TODAY()-I2<=30),5,0). Timing in N: =IFS(TODAY()-H2<=2,25,TODAY()-H2<=7,15,TODAY()-H2<=30,5,TRUE,0). With dates only, "two days or less" stands in for 48 hours.
Apply penalties and exclusions last
Total in O: =IF(K2="Y","Exclude",L2+M2+N2-IF(J2="Y",30,0)). Excluded rows return text instead of a number, so they cannot be sorted into a sending list by accident.
Write the reasons with a formula
Reasons in P: =TEXTJOIN("; ",TRUE,IF(B2="Y","target industry",""),IF(C2="Y","size in band","size outside band"),IF(D2="Y","decision-maker verified","no verified email"),G2&" on "&TEXT(H2,"mmm d")). Keep the source link in its own column so anyone can open the evidence in one click.
Sort every morning
Sort by column O, filter out "Exclude" and work from the top. Because the timing formula uses today's date, rankings change as signals age even when nothing else does.
How do you explain a score with reasons?
A bare number invites two bad reactions: blind trust or blind dismissal. Write the reasons next to every score, one line per part, with a link to the evidence. For lead A above, it reads:
- Fit (40): B2B SaaS, 30 people, United States; founder identified, business email verified.
- Intent (30): asked for cold email tool recommendations in a public thread (link to the post).
- Timing (25): posted 19 hours ago.
Why reasons matter
Reasons do three jobs. Reps can check them in seconds and write a better first line from them: the intent line usually becomes the opening sentence, as in our cold email templates. Anyone can spot a wrong reason ("agency" when the company is a SaaS product) and fix the rule that produced it. And when a prospect asks why you contacted them, you have a clear, honest answer.
If you score people rather than only companies, the data is personal data. California residents, for example, can ask what you hold about them under the CCPA. Keep only what the score needs, and be ready to show it.
This is why Startories writes out the reasons behind every lead score instead of showing a bare number: you see the fit, the signal and its link, and you can correct the targeting. More on that in AI lead generation.
How do you review and recalibrate the model?
Review a sample every week
Read 20 leads above the threshold and 10 just below it. Note every case where you disagree with the score, and why.
Track outcomes by score band
For each band (70 and above, 50 to 69, 30 to 49), record contacts, replies, positive replies, meetings held and opportunities. Wait for a few hundred contacts before drawing conclusions.
Check that the bands separate
Higher bands should produce more positive replies and meetings per contact. If 50 to 69 performs like 70 and above, your threshold is too high; if a criterion shows up equally in good and bad outcomes, it is worth too many points.
Change one thing at a time
Adjust a few criteria, write down what changed and when, and compare the next period with the last one.
Re-check after big changes
A new offer, a new segment or a new signal source can make old points meaningless. Re-score a sample by hand after each one.
| Band | Contacts | Positive replies | Meetings held | Meetings per 100 contacts |
|---|---|---|---|---|
| 70 and above | 180 | 14 | 9 | 5.0 |
| 50 to 69 | 310 | 15 | 8 | 2.6 |
| 30 to 49 (small test sample) | 100 | 2 | 1 | 1.0 |
How to read the example
The bands come out in the right order, so the model ranks something real. The 50 to 69 band still books meetings at about half the rate of the top band; whether it deserves your sending capacity depends on what a meeting is worth to you. The test sample from the band you normally skip confirms that the 50-point line is roughly in the right place.
Next, break the meetings down by criterion. Suppose six of the eight meetings in the 50 to 69 band came from leads whose only intent was a related job post. That suggests hiring signals are worth more in your market than the 15 points a generic company event gets. Give hiring its own row on the Points tab at 20 and compare next quarter.
Be careful with small numbers. One meeting more or less moves the test sample from 1.0 to 0 or 2.0. Treat a difference of a few meetings as noise until the totals grow.
What are the most common lead scoring mistakes?
- Scoring email opens. Apple Mail Privacy Protection and similar features load images in the background, so opens are recorded whether or not anyone read the email. Score replies, not opens.
- Counting one event twice. A funding round, the press release about it and the founder's post announcing it are one signal, not three. The second-signal bonus is for independent events.
- Too many criteria. A model with 30 rules is impossible to debug. Ten to fifteen criteria cover most outbound needs.
- No negative points. Without penalties and exclusions, a strong signal from a company you cannot serve still rises to the top.
- Scores without reasons. Nobody trusts, or fixes, a number they cannot explain.
- Frozen timing points. A score computed once and never refreshed keeps a month-old signal looking urgent.
- Never recalibrating. Markets, offers and sources change. A model nobody has touched in six months is likely wrong somewhere.
- Treating the score as qualification. A high score earns a well-researched first message, not a place in your forecast.
How to start scoring your outbound leads
Write your profile first, then copy the sample model into a spreadsheet, score 50 recent leads by hand and compare the ranking with your own judgment. Adjust until you agree with most of it, then apply it to every new lead and review it weekly. Our outbound sales strategy guide shows where scoring fits in the full process, and teams that build their own enrichment and scoring workflows can weigh the trade-offs in Clay vs Apollo.
If you would rather not maintain it by hand, Startories scores every signal-based lead against your profile with the reasons written out, verifies the decision-maker and drafts the outreach. Plans start at $99 a month: see pricing. If you want the team to set up the scoring, targeting and sending for you, look at the done-for-you option.
Frequently asked questions
What is a good lead score threshold?
There is no universal number. Start where your own review agrees with the ranking, for example 70 out of 100 in the sample model, then move it after a few hundred contacts based on which score bands actually produce positive replies and meetings.
What is the difference between lead scoring and lead qualification?
Scoring ranks prospects before you talk to them, using data you can observe. Qualification happens in the conversation, when you confirm need, budget, authority and timing. A high score earns a good first message; only qualification earns a place in the pipeline.
What is negative lead scoring?
Negative scoring subtracts points for traits that make a purchase less likely, such as a company outside your size band or a personal rather than business need. It keeps strong signals from poor-fit companies off the top of your list. Hard disqualifiers belong in exclusions instead.
Should I use predictive or AI lead scoring?
Predictive models need many historical outcomes to learn from, which young companies rarely have. Start with a transparent rules-based model, collect outcomes, and consider predictive scoring later, as long as it still shows the reasons behind each score.
How often should I update a lead scoring model?
Review a sample weekly and recalibrate monthly or after a few hundred contacts. Also re-check it whenever you change your offer, your target segment or your signal sources, because old points may no longer predict anything.
Can I do lead scoring without a CRM?
Yes. A spreadsheet with one column per criterion, a total and a reasons column works well for the first few hundred leads. Move it into your CRM or another tool once the model is stable and you trust the ranking.