# The Signup-to-Account Queue: Rolling Scattered PLG Signups Into Ranked PQAs

*You will turn a raw signup export into an account-level, ranked PQA queue for AEs, with explicit do-not-touch rules that protect self-serve conversion.*

- Canonical URL: https://www.refolk.ai/guides/signup-to-account-queue
- Pillar: Sales and go-to-market
- Format: Playbook
- Published: 2026-10-05
- Last reviewed: 2026-10-05
- Reading time: 16 min

This is the procedure for turning this week's flood of individual free-product signups into a deduplicated, company-level list of accounts worth a sales touch. It is for founders selling their own product, account executive and SDR leads, and RevOps teams running a product-led motion. By the end you can take a raw signup export and hand AEs a ranked product-qualified-account queue, with explicit do-not-touch rules that protect the signups converting on their own.

Most product-qualified-lead content stops at scoring one user. The hard parts live on either side: resolving scattered individuals into a single buying unit, and then deciding which units to leave alone. This guide covers both.

## Why you cannot route every signup

Routing every free signup to sales wastes most of the effort, because the base rate of conversion is too low to justify undifferentiated outreach. Published freemium free-to-paid rates sit in low single digits, so a queue of raw signups is overwhelmingly made of people who will never buy, whether or not a human calls them.

The numbers are consistent across independent datasets. First Page Sage, drawing on 80+ SaaS clients between 2021 and 2025, found an average freemium-to-paid conversion of 3.7%. The ChartMogul and ProductLed report calls 3-5% a good self-serve freemium rate and 8-12% great. OpenView's often-cited range puts the typical rate at 2-5%, with the full range from 1% to 10%.

| Source | Good | Great / typical | Basis |
|---|---|---|---|
| First Page Sage | - | 3.7% avg | 80+ SaaS clients 2021-25 |
| ChartMogul / ProductLed | 3-5% | 8-12% | 200 products |
| OpenView (via Artisan) | 2-5% | 1-10% full range | OpenView benchmarks |

The implication is blunt. At a 2-5% typical rate, routing all signups means roughly 95% of AE effort hits non-buyers. The benchmark does not tell you how to score a lead; it tells you that scoring is not optional. Triage comes first, outreach second.

**2-5% - Typical freemium free-to-paid conversion rate**

OpenView's cited range. At this base rate, routing every signup sends AEs mostly to non-buyers.

One caveat worth keeping in view: free trials behave differently from freemium. A 2025 dataset of 10,000+ companies and 2.5M trial users put median B2B trial-to-paid at 18.5%, far above freemium's low single digits. If your motion is a time-boxed trial rather than an always-free tier, your base rate is higher and your triage can be looser. The procedure below is the same either way; only your cutoffs move.

## The real problem: signups are people, accounts are the unit

PLG signups arrive as individuals, not accounts, and that mismatch is where routing quietly breaks. A VP of Engineering signs up with a personal email, a developer on the same team signs up with a company email, and your CRM now holds three disconnected leads from one company. Without lead-to-account matching, routing treats them as three separate strangers and reaches out three times.

This is the step most PQL guides skip, and it is the largest single lever on AE time. B2B buying is organizational: five colleagues using the product are one opportunity, not five leads. A queue of 500 individual signups may be 120 accounts once you roll it up. Deduplication alone can cut the apparent volume by three-quarters before anyone scores a thing.

The resolution method has a fixed order. Match on email domain first, because in B2B the domain is often the strongest account signal. Then handle the cases the domain cannot: personal domains like Gmail and Outlook produce no usable domain signal, large enterprises may run multiple domains, and shared domains create false positives. For personal emails the right default is flag, don't block: accept the signup, tag it `email_type = personal`, then find the person's work address from their name and the company they typed at signup.

Every enriched identity must carry a confidence score. For shared or ambiguous domains such as consultants, agencies, and Gmail signups, flag low-confidence identifications rather than enriching incorrectly. A confident but wrong title is more dangerous than an honest unknown, because it will route someone who was never there.

> **Rule:** Resolve to account before you score
>
> Scoring an individual signup before rolling it up to its account produces duplicate outreach to one buying committee. The account is the unit of the decision; the person is only an input.

#### From raw signups to a routed queue

| Stage | Figure | Note |
| --- | --- | --- |
| Raw signups | 500 | one row per user |
| Deduped users | 460 | exact-duplicate emails removed |
| Resolved accounts | 120 | users rolled up by domain |
| Scored PQAs | 30 | met value, fit, and intent |
| Tier 1 routed | 8 | same-day human outreach |

*Volume narrows at every stage, and the account rollup is the steepest drop.*

The funnel figures are illustrative of the shape, not a benchmark; your own ratios depend on your product and signup mix. What is not illustrative is the direction. Each stage removes work, and the largest removal is the rollup, not the score.

## Deriving the 3-5 behaviors that actually predict conversion

There is no universal definition of a qualified signup; the definition lives in your data, and you find it by comparing two cohorts. Pull your last 50-100 conversions and an equal number of accounts that signed up but did not convert, find the actions that separate them, add a time dimension, and write each as a sentence you can test backwards.

The two-cohort discipline matters because converters alone lie by omission. An analysis of converters only tells you that paying customers use your product, which you already knew. The signal is in the difference. A behavior becomes a definition by its lift: "invited a second user in week one," done by 70% of converters and 12% of non-converters, is a definition. A behavior both groups do at similar rates is noise dressed as insight.

The size of the effect you are hunting for is real. Users who activate certain features within specific timeframes convert at 3-5x higher rates than those who do not. You are looking for three to five such behaviors, each with the converter-versus-non gap written down, so that anyone can audit why a behavior is in the model.

**PQL behavior definition, written as a testable sentence**

```
Behavior: <action> within <time window>
Converter rate: __% did this
Non-converter rate: __% did this
Lift: converters are __x more likely
Verdict: KEEP if the gap is large and the action is early; DROP if both cohorts do it equally
```

*Fill in your own action, window, and the two cohort percentages. One line per behavior, three to five lines total.*

Add the time dimension deliberately. "Invited a second user" is weaker than "invited a second user in week one," because the window is what makes the signal early enough to act on. Best-in-class B2B PLG products reach time-to-value in under 10 minutes; north of an hour means the product is broken for self-serve. If your meaningful actions cluster in the first session, your windows should be tight.

Ask me this: `List growth analysts or product analysts who have built PQL or PQA scoring models from cohort data.` - [run the search](https://www.refolk.ai/start?q=List%20growth%20analysts%20or%20product%20analysts%20who%20have%20built%20PQL%20or%20PQA%20scoring%20models%20from%20cohort%20data.).

*Returns people who have run the two-cohort lift analysis before, so you are not inventing the method from scratch.*

Most teams building scoring have never done this. In [Refolk](/)'s index there are 2,376 US professionals skilled in lead scoring but only 1,271 skilled in product-led growth. The people who can build a model mostly think in MQL terms of fit and intent, not in product behavior and lift. That gap is worth naming before you hand the work to someone, because a lead-scoring background without a PLG instinct will quietly rebuild an MQL model on signup data.

| Segment | Count | Derived ratio |
|---|---|---|
| Product-Led Growth skill, US | 1,271 | baseline |
| Product-Led Growth skill, UK | 265 | US is 4.8x UK |
| Lead Scoring skill, US | 2,376 | 1.87x the PLG-US pool |

UK teams are structurally thinner here. With the US pool 4.8x the UK pool, a UK founder is far more likely to be inventing this process without in-house precedent. If that is you, lean harder on the written definitions; they are what lets one person carry the method.

## The procedure, end to end

The method runs in nine steps, from raw export to a calibrated queue. Steps 1 through 4 resolve signups into accounts, step 5 defines what qualifies, steps 6 through 8 score and route, and step 9 keeps it honest over time. One row of your source data should be traceable through every stage.

#### Signup export to ranked PQA queue

1. **Export and dedupe raw signups** - Pull the week's signups with email, timestamp, typed company, and first product events into one sheet, then remove exact-duplicate emails. Done when you have one row per user.
2. **Classify email type** - Tag each row as business or personal against a freemail list. Flag, do not block: accept and set email_type = personal. Done when every row carries an email_type.
3. **Resolve to company** - Match on email domain first, then enrich personal domains from name plus typed company with a confidence score, flagging low-confidence matches rather than enriching wrong. Done when each user maps to an account ID with a confidence value.
4. **Roll users up into accounts** - Group all resolved users under one account and merge multiple workspaces from the same domain. Done when the three-people-from-one-company case is a single account row.
5. **Derive PQL behaviors from cohorts** - Pull 50-100 conversions and an equal non-converter cohort, find the separating actions, add a time dimension, and write each as a sentence. Done when you have 3-5 behaviors with lift documented.
6. **Score accounts on value, fit, and intent** - Require all three to be met, then apply your score bands or trigger tiers at the account level. Done when each account has a composite score.
7. **Apply do-not-touch rules** - Suppress accounts below threshold, in active self-serve conversion, or in churn-risk, routing declining accounts to CS. Done when you have a held-back list with a reason code per account.
8. **Rank and route** - Map score bands to actions, attach usage context to each alert, and flag Tier 1 for same-day touch. Done when AEs have a ranked PQA queue in hand.
9. **Calibrate weekly or monthly** - Tune cutoffs from SQO and win-rate feedback and log the false-positive rate. Done when thresholds are adjusted and the drift is recorded.

Steps 1 and 2 are automatable in minutes once set up; step 3 runs in batches through an enrichment tool; step 5 is a one-to-two-day analysis you repeat quarterly, not weekly. The rest you run every time a batch of signups lands.

## Scoring and routing: two frameworks, one decision

Score accounts on value realization, customer fit, and buying intent together, requiring all three, then map the result to a routing action. Two published frameworks give you the cutoffs: a score-band model and a trigger-rate model. They are alternative engines for the same step, not sequential stages.

| Band / tier | Score model | Trigger-rate model | Action |
|---|---|---|---|
| Hot | 80-100 | >30% | Same-day human outreach |
| Warm | 60-79 | 15-30% | Automated outreach / 72h |
| Nurture | 40-59 | 10-15% | Product-led nurture |
| No touch | <40 | <10% | Self-serve only |

The score-band model assigns composite points and routes the top band to immediate sales outreach, the next to automated outreach plus a CS check-in, the middle to product-led nurture, and the bottom to self-serve with no sales touch. The trigger-rate model is more empirical: it measures the actual conversion rate that follows a given signal, routing triggers above 30% to same-day human touch and anything below 10% to automated nurture only.

Pick the trigger-rate model if you have enough historical volume to measure conversion rates per signal reliably; pick the score-band model if you are earlier and need a composite proxy. Either way, set cutoffs empirically. Back-test the top 10-15% of historical conversions to find where the line sits, then tune monthly from SQO and win-rate feedback. A cutoff picked by feel will either flood AEs or starve them.

Attach usage context to every alert. A score with no story behind it gives the AE nothing to open with. The alert should say what the account did, not just that it crossed a line.

> **Watch out:** A score that triggers nothing is analytics, not automation
>
> Every band must map to a workflow with an owner and an SLA. If your "hot" accounts land in a dashboard nobody actions, you have built a report, not a routing system.

## How this goes wrong

The failure modes here are specific and mostly silent: they do not throw errors, they just route the wrong accounts or burn the right ones. Each has a tell and a check. Walk this list before you trust a queue.

### Resolution and enrichment failures

**Domain match false positive.** Shared domains from agencies and consultancies, or multi-brand enterprises, collapse unrelated users into one account. The tell is an account whose users span unrelated typed-company names. Check by flagging any such account and requiring a confidence score before merge.

**Personal-email over-enrichment.** Enriching a Gmail signup to the wrong person. It looks like a confident title and company the user never typed. Check by flagging low-confidence identifications rather than enriching incorrectly; carry the confidence score through to routing.

### Scoring failures

**Converter-only behavior analysis.** Deriving rules from winners alone tells you paying customers use your product, which you knew. Check by requiring an equal non-converter cohort and computing lift for every behavior.

**Firmographic-only trigger.** The most common PLG trigger mistake is treating "large company signed up" as a trigger. Firmographic signals without behavioral signals produce cold outreach to a logo that has done nothing. Check by refusing to route any account without a behavior event.

**Power-user false positive.** A single power user who has explored every feature for six months is deeply engaged and may never convert. Check by requiring account-level breadth, meaning multiple active users, not depth from one.

### Restraint failures

**Poaching self-serve conversions.** Sales touches an account already converting on its own. Manually prospecting into self-serve accounts that have not crossed the threshold is the single fastest way to poison the PLG funnel. Check by suppressing active-checkout and recent-upgrade accounts with a reason code.

**Expansion outreach into churn risk.** If an account's health score is below threshold, from low usage, declining seats, or a recent downgrade, expansion outreach is counterproductive. Check by routing declining accounts to customer success for retention, not to AEs.

> The product sells the first seat. Sales earns its keep on the accounts the product cannot finish alone.

## The do-not-touch rules

The suppression list is the part that protects your self-serve motion, and it is the rarest discipline in practice. Only 24-35% of companies implement PQL scoring at all, so the restraint layer on top of it is rarer still. Three categories of account get held back, each with a reason code so the suppression is auditable rather than invisible.

#### Touch or hold, by conversion state and health

Horizontal axis runs from Not self-converting to Actively self-converting. Vertical axis runs from Healthy / growing to Declining / at-risk.

| Quadrant | What it means |
| --- | --- |
| Route to AE | Crossed the threshold and not converting alone; this is the PQA queue |
| Hold, self-serve | Converting on its own; touching it risks the deal and the funnel |
| Route to CS | At-risk but not converting; retention before any expansion talk |
| Hold, CS watch | At-risk and mid-conversion; let CS steady it, keep AEs out |

*The two axes that decide whether an account is worth a human, or should be left to convert on its own.*

The first category is accounts below your routing threshold. Sales should not manually prospect into self-serve accounts that have not crossed the line. The product targets these users; sales takes over only after they qualify. The second is accounts in active self-serve conversion, such as a recent self-serve upgrade or an active checkout. These are converting without you, and a sales touch introduces friction and discount pressure where none was needed. The third is churn-risk accounts, which go to CS for retention, not to AEs for expansion.

The active-conversion window is this guide's own contribution. Public sources establish the threshold rule and the churn-risk rule plainly; they do not publish a named ruleset for "active checkout" or "recent self-serve upgrade." I present those as grounded in the same logic: an account the product is about to close itself should not be interrupted. Treat the specific window as a starting hypothesis and confirm it against your own data, by checking whether sales touches during self-serve conversion correlate with lower contract value or higher discounting.

> **Tip:** Automate the handoff, do not leave it to judgment
>
> Define the rule for when the product owns the user and when sales takes over, then automate the handoff, for example after a user dismisses the last in-product prompt. A rule enforced by a system holds; a rule enforced by goodwill does not.

## Calibration and the pre-launch check

Thresholds drift, so the queue is never finished; it is tuned. Recalibrate monthly from SQO and win-rate feedback, and log your false-positive rate each cycle so you can see the model degrading before AEs start complaining. The behavior derivation itself should be re-run quarterly, because what separated converters two quarters ago may not hold after a product change.

Before you hand the first queue to AEs, confirm the work with this list. Each item is a thing to verify, not a topic to think about.

#### Before this queue reaches an AE

- [ ] Every user row carries an email_type of business or personal.
- [ ] Every resolved identity carries a confidence score, and low-confidence matches are flagged not merged.
- [ ] The three-people-from-one-company case appears as a single account row.
- [ ] Each PQL behavior has a documented converter-versus-non-converter lift.
- [ ] No account is routed on firmographics alone, without a behavior event.
- [ ] Every score band maps to a workflow with an owner and an SLA.
- [ ] Accounts in active self-serve conversion are suppressed with a reason code.
- [ ] Declining and at-risk accounts are routed to CS, not AEs.
- [ ] Tier 1 accounts carry usage context, not just a number.

When those pass, the queue is ready. It is a ranked list of accounts, deduplicated from scattered individuals, each either routed with a reason and context or held back with a reason code. That is the deliverable: AEs spend their hours on the accounts that need a human, and the ones converting on their own are left to do so. Re-run the whole procedure on each fresh batch of signups, and let step 9 keep the cutoffs honest as your product and your signup mix move.

## Frequently asked questions

### How many signups should I actually route to sales each week?

Far fewer than you think. With typical freemium conversion at 2-5%, routing everyone wastes most AE time on non-buyers. Published frameworks route only the top band: scores of 80-100 or triggers with a conversion rate above 30% get same-day human outreach. Back-test the top 10-15% of your historical conversions to set the cutoff, then tune it monthly. Expect your routed queue to be a small fraction of raw signups once you roll up to accounts.

### What do I do with Gmail and other personal-email signups?

Flag, do not block. Accept the signup and tag it email_type = personal, then try to resolve the person's work address from their name plus the company they typed at signup. Attach a confidence score to every enriched identity. For shared or ambiguous domains like agencies and consultancies, flag low-confidence matches rather than enriching incorrectly, because a confident but wrong title is worse than an honest unknown.

### How is a PQA different from a PQL and which one do I route?

A PQL qualifies on one person's in-product behavior; a PQA qualifies on an account's aggregate behavior across multiple users, team invites, and breadth of adoption. In B2B, five colleagues using the product are one opportunity, not five leads, so you route the account-level PQA to AEs for sales-assist and expansion. In SMB where the user is the buyer, a PQL can be sold to directly.

### Why can't I just use converters to define what a good signup looks like?

Because an analysis of converters alone only tells you that paying customers use your product, which you already knew. You need an equal cohort of accounts that signed up and did not convert, then compute the lift between them. A behavior like "invited a second user in week one" done by 70% of converters but only 12% of non-converters is a definition; a behavior both groups do equally is noise.

### Which accounts should sales never touch, even if they look hot?

Three groups. Accounts below your routing threshold, which sales should not manually prospect into because it is the fastest way to poison the funnel. Accounts in active self-serve conversion, such as recent upgrades or active checkout, which are converting on their own. And churn-risk accounts with low usage, declining seats, or a recent downgrade, which go to customer success for retention, not to AEs for expansion.

---

*From the Refolk guide library. I revise these guides rather than replacing them, so the current version is always at https://www.refolk.ai/guides/signup-to-account-queue*