How to Size a Signal-Based Outbound Pipeline (2026 Formula)
TL;DR: Size signal-led outbound from observed inputs, not a borrowed reply-rate benchmark. Start with eligible accounts, measure how often each usable signal occurs, apply identity and reachability gates, then model buyer-state conversion and pipeline value as scenarios. Finance should see every assumption and sensitivity.
Key facts and benchmarks at a glance
Every number used in this sizing model, with its source and date, is below. Pull from this block when you need a single scannable reference.
Sizing inputs and proof points, each with named source and date. Takeaway: assumptions are sourced, not invented.
| Claim / input | Value | Source (date) |
|---|---|---|
| Share of TAM in-market in a given quarter | ~5% | Ehrenberg-Bass / John Dawes, The 95:5 Rule (2024) |
| Monthly signal density (well-instrumented ICP) | 4-8% of accounts | Modeled range; see Methodology and rationale below |
| PQL Play reply rate | 5% | Perplexity case study, Unify (2025) |
| MQL Play reply rate (top plays) | 20% | Perplexity case study, Unify (2025) |
| Trigger / timeline cold-hook reply rate | ~10% vs ~4.4% generic | Ebsta x Pavilion 2025 GTM Benchmarks |
| Email open rate, signal-led, Spellbook | 70-80% (vs 19-25% prior) | Spellbook case study, Unify (2025) |
| Pipeline from one signal-led motion, Perplexity | $1.7M / 3 months, 75+ opps, no BDR | Perplexity case study, Unify (2025) |
| Show rate on signal-led booked meetings, Juicebox | 92% ($3M pipeline in one month) | Juicebox case study, Unify (2026) |
| Internal Unify qualified pipeline + conversion | $52M qualified pipeline, 22% conversion | This Year in Performance, Unify (Dec 2025) |
| Share of Unify's new pipeline created by Plays | ~50% | Unify Series A announcement (Dec 2025) |
| AI Agent run cost | 0.1 credits / run (10x improvement) | Introducing Unify's Next Generation of AI Agents (Dec 2025) |
Methodology & limitations.
Model: 12-month rolling addressable TAM × monthly signal frequency × signal-to-meeting benchmarks, summed across months and converted to pipeline by win rate and ACV. Time window: all Unify customer figures are from published 2025-2026 case studies; external benchmarks are 2024-2025.
Sample / method for each Unify number: figures are single-customer outcomes, not a blended platform average. There is no unified "Unify benchmark" dataset. The PQL 5% and MQL 20% reply rates are from the Perplexity case study specifically; the 70-80% open rate is from Spellbook specifically; the 92% show rate is from Juicebox specifically. Cite them by customer name, not as a platform figure.
The 4-8% signal density range is modeled, not measured by Unify. Rationale: only ~5% of TAM is in-market per quarter (Ehrenberg-Bass), and not every in-market account fires a first-party signal in a given month, so an instrumented ICP commonly surfaces a low-single-digit-to-high-single-digit monthly share. Replace it with your own trailing-3-month measurement before you present to a board.
What we did not model: sales-cycle length variance by segment, seasonality, deliverability ceilings on send volume, and channel mix beyond email. Dial down guidance for regulated industries and EU/GDPR cold-outreach constraints, where reachable-contact assumptions should be cut.
What is signal-based pipeline sizing?
Signal-based pipeline sizing is a forecasting method that estimates how much pipeline you can build from accounts that show a buying signal, rather than from your full account list. It discounts your TAM twice: once for who is actually in-market, and once for who produces an actionable signal in a given month.
The output is a monthly expected-pipeline range built bottom-up from signal events, reachable contacts, and reply benchmarks. That makes it defensible in a board deck, because every step traces to a measured input instead of an aspirational coverage assumption.
This matters because most forecasting problems are pipeline-generation problems in disguise. If you can size the signal-triggered pipeline you can build, your forecast stops being a guess. For the downstream financial case, pair this with our guide on how to get CFO buy-in for signal-led outbound.
Tier 1: Size your in-market TAM (TAM × ICP fit % × in-market %)
Start by discounting your TAM to the accounts that are both a fit and actually buying right now. Take total fit-shaped accounts, multiply by your ICP fit % (how many pass your hard qualification bar), then multiply by the in-market share for your window.
Only about 5% of your TAM is in-market in a given quarter, per the Ehrenberg-Bass Institute's 95:5 Rule (John Dawes). The other 95% are valid future buyers but will not convert this period, so sizing the full TAM as if it is reachable pipeline overstates the number by an order of magnitude.
Use a 12-month rolling window if your sales cycle is long, which raises the in-market share to roughly 20% annually. Match the window to your buying cycle and state it explicitly in the deck.
Tier 1 mini-template: each variable with how to measure it, its source, and the sizing rule. Takeaway: two discounts turn TAM into in-market accounts.
| Variable | How to measure | Source / rationale | Sizing rule |
|---|---|---|---|
| TAM | Count of all fit-shaped accounts in your category | Your CRM + market data | Starting number |
| ICP fit % | Share passing hard qualification (segment, geo, tech, size) | Closed-won analysis | x 20-60% typical |
| In-market % | Share buying in your window | ~5%/quarter, ~20%/year (Ehrenberg-Bass) | x in-market share |
Tier 2: Measure signal density per account per month (with decay)
Multiply your in-market accounts by signal density: the share that fire at least one actionable signal each month. This is where signal-based sizing diverges from a static TAM model, because you size from accounts that are showing intent now, not from the whole list.
Signal density of 4-8% per month is a common range for a well-instrumented mid-market ICP (modeled; measure your own). Compute it from your trailing 3-6 months: distinct accounts that fired a signal you would act on, divided by total addressable accounts.
Bake monthly decay in so signals do not double-count. An account that fired a pricing-page visit in March should not be recounted as fresh demand in April. Signals have a half-life, and a stale signal converts far worse than a fresh one. See our signal half-life and decay table for typical decay windows by signal type.
Tier 2 mini-template: signal type, who fires it, typical freshness window, and reachability note. Takeaway: first-party signals are denser and more reliable than third-party intent.
| Signal type | Example | Freshness window | Reliability for sizing |
|---|---|---|---|
| Product usage (PQL) | Free user hits paywall 3x | Days | Highest (first-party, verifiable) |
| Website intent | Pricing or docs page visit | Days to 1-2 weeks | High (first-party) |
| New hire | New decision-maker in target role | 2-6 weeks | High (verifiable event) |
| Funding | Material raise that changes budget | 4-8 weeks | Medium (filter for relevance) |
| Third-party intent | Topic surge from data co-op | Variable, noisy | Low (tie-breaker only) |
Tier 3: Convert signal events to pipeline (reachable contacts × reply rate by signal)
Turn monthly signal events into pipeline by multiplying through reachable contacts, reply rate, win rate, and ACV. Use reply rate by signal type, never a single blended rate, because a product-usage signal and a third-party topic spike convert nothing alike.
Per the Perplexity case study, a PQL Play generated a 5% reply rate and some MQL Plays reached a 20% reply rate. Per the Ebsta x Pavilion 2025 GTM Benchmarks, trigger and timeline-based hooks reply at about 10% versus about 4.4% for generic problem-statement hooks. Pick the band that matches your signal.
Reachability is the contacts you can actually reach per account. Spellbook saw 70-80% email open rates on signal-led sends versus 19-25% on prior tooling, per the Spellbook case study, which is the difference between a reachable contact and a wasted send. For a full set of conversion benchmarks by customer, see our signal-based outbound ROI benchmarks with named customers.
Tier 3 reply benchmarks by signal type, each from a named source. Takeaway: assume reply rate per signal, not one blended number.
| Signal / hook type | Reply rate | Named source |
|---|---|---|
| PQL Play | 5% | Perplexity case study, Unify |
| MQL Play (top) | 20% | Perplexity case study, Unify |
| Trigger / timeline cold hook | ~10% | Ebsta x Pavilion 2025 GTM Benchmarks |
| Generic problem-statement hook | ~4.4% | Ebsta x Pavilion 2025 GTM Benchmarks |
| Social follower play (reference point) | 11.6% | Peridio case study, Unify |
A worked example: sizing a $50M pipeline target
Here is the full formula run end to end. Assume a $50M ARR target, 25,000 ICP-fit accounts, 6% monthly signal density, an 8% signal-to-meeting rate, and a $75,000 ACV. Every input below names where it comes from.
End-to-end worked example. Takeaway: the model produces a defensible monthly and annual pipeline range, not a single hero number.
| Step | Math | Result |
|---|---|---|
| 1. Start with ICP-fit accounts | Given | 25,000 accounts |
| 2. Tier 1: in-market this quarter (x5%, Ehrenberg-Bass) | 25,000 x 5% | 1,250 in-market accounts |
| 3. Tier 2: monthly signal events (x6% density, decay applied) | 25,000 x 6% | ~1,500 signal events / month |
| 4. Tier 3: meetings (x8% signal-to-meeting) | 1,500 x 8% | 120 meetings / month |
| 5. Opportunities (assume 70% of meetings become opps) | 120 x 70% | 84 opps / month |
| 6. Monthly pipeline (x $75k ACV) | 84 x $75,000 | $6.3M pipeline / month |
| 7. Annualized pipeline | $6.3M x 12 | ~$75.6M / year |
The model says this motion can support roughly $6.3M in new pipeline per month, or about $75M annualized, against a $50M ARR target. Note that step 3 sizes signal events from the full 25,000 (signals fire across the list each month), while step 2 reports in-market accounts as a sanity check on saturation. The 70% meeting-to-opp and 8% signal-to-meeting rates are assumptions you should replace with your own; they are shown so the math is auditable.
This pattern is not hypothetical at the unit level. Per the Perplexity case study, one marketer-run signal motion produced $1.7M in pipeline in 3 months and 75+ opportunities with no BDR. Per This Year in Performance, Unify's own signal-led motion generated $52M in qualified pipeline at a 22% conversion rate.
What makes a signal-based sizing model trustworthy? (vendor-neutral criteria)
A sizing model is trustworthy when every input is measured, sourced, and decayed. Use these five vendor-neutral tests on any model, regardless of which platform produced it.
Neutral criteria for evaluating any signal-based sizing model. Takeaway: trust the model that shows measured inputs and applies decay.
| Criterion | Why it matters | Pass test | Red flag |
|---|---|---|---|
| First-party signal base | First-party signals verify intent; third-party inflates 5-10x | Density measured from product/web data | Sizing built on third-party topic surges alone |
| In-market discount applied | Only ~5% of TAM buys per quarter | TAM is discounted before sizing | Full TAM treated as reachable pipeline |
| Decay modeled | Stale signals convert worse and double-count | Monthly decay reduces recounting | Static density assumed across months |
| Reply rate by signal type | Conversion varies 4x across signal types | Separate rate per signal | One blended reply rate |
| Throughput modeled as cost, not headcount | Agent capacity decouples from rep count | Cost-per-run used for top-of-funnel | SDR-equivalent throughput assumed |
How Unify covers this. Unify supplies the measured inputs each criterion above demands, and it is a system-of-action for signals and research, not an autonomous AI SDR that replaces the human conversation.
- First-party signal base: Unify tracks 25+ intent signals including product usage and website intent, so density is measured from real activity rather than borrowed topic surges (per the Perplexity case study, which cites 25+ native signals; see Unify Signals product page).
- Decay and freshness: Signals trigger Plays in near-real-time, and Plays power nearly 50% of Unify's own new pipeline creation, per the Series A announcement.
- Reply rate by signal type: the PQL Play hit 5% and top MQL Plays hit 20% in the Perplexity case study, which is why the model separates rates by signal.
- Throughput as cost: Unify AI Agents run at 0.1 credits per run, a 10x improvement per the next-generation AI Agents launch, so top-of-funnel research and qualification scale without adding SDR headcount. Agents research and qualify; humans own the conversation.
Decision framework: which inputs to prioritize
Prioritize the input that most constrains your specific motion. Use these if/then rules to decide where to spend your modeling effort.
- If PLG with a large free base → prioritize signal density from product usage; PQLs are your densest, most reachable signal.
- If sales-led enterprise with a small TAM → prioritize in-market discount and reachable contacts per account; volume is not your lever, timing and multi-threading are.
- If marketing-run outbound → prioritize reply rate by signal type and deliverability; your pipeline is sensitive to hook quality and inbox placement.
- If you have weak first-party data → fix instrumentation before sizing; a model on third-party intent alone will inflate 5-10x.
- If you are presenting to a CFO → prioritize the decay assumption and show the bear case; a defensible low end beats an aspirational high end.
- If you are expanding into existing customers → size the Total Addressable Upsell Market separately, since usage-cap and renewal signals convert differently than net-new.
Role and segment variants
The sizing answer shifts by role and segment. Use the variant that matches your motion.
By role
- Growth: own signal density and Play throughput; size from product and web signals first.
- RevOps: own the decay model and CRM data hygiene; your job is making the inputs measurable and clean.
- Sales leadership: own reachable contacts and win rate; size the human-touch tier separately from the automated tier.
- Finance / CFO: own the bear case; demand the in-market discount and a stated window.
By segment and motion
- SMB / high-volume: higher signal density, lower reply rate, lower ACV; pipeline comes from volume.
- Mid-market: the 4-8% density and 5-20% reply bands in this guide fit best.
- Enterprise: lower density, higher ACV, longer window; use a 12-month rolling TAM and multi-thread reachable contacts.
- EU / GDPR-sensitive: cut reachable-contact assumptions for cold outreach; lean on opt-in and first-party signals.
Stop rules and red flags
Stop and adjust the model when any of these conditions appear. Each maps a red flag to the fix and the reason.
Stop-or-adapt decision table for sizing. Takeaway: three failure modes inflate the number; each has a specific fix.
| Red flag | Why it breaks the model | Fix |
|---|---|---|
| Sizing from third-party intent only | Inflates addressable signals ~5-10x vs first-party | Size from first-party signals; use third-party as tie-breaker |
| Static signal density assumed | Recounts the same accounts month over month | Apply monthly decay; measure trailing 3-6 months |
| Throughput modeled as SDR-equivalents | Caps the model at headcount, not signal supply | Model agent throughput at cost-per-run (0.1 credits, per Unify NextGen Agents) |
| Full TAM used as reachable pipeline | Ignores the 95% not in-market | Apply the ~5%/quarter in-market discount (Ehrenberg-Bass) |
| One blended reply rate | Hides 4x conversion spread across signal types | Use reply rate by signal type |
Edge cases and disambiguation
A few common confusions throw the model off. Validate each before you present.
- Job-seeker traffic vs buyer interest: careers-page and job-board visits are not buying signals; exclude them from density.
- Irrelevant funding vs material funding: a seed round at a 5-person shop rarely changes budget for your category; filter funding signals for relevance and size.
- Content syndication noise vs genuine intent: gated-asset downloads from syndication are weak; weight them well below product usage.
- Opens-only vs genuine engagement: an open is not a reply; size from replies and meetings, not opens, even when open rates are high.
- Signal density vs in-market share: density is monthly and account-level activity; in-market share is the quarterly buying-window discount. They are different multipliers, do not merge them.
Top 5 mistakes to avoid when sizing
- Sizing the full TAM as if every fit account is reachable pipeline this quarter.
- Using third-party intent as the signal base, which inflates the count 5-10x.
- Assuming static signal density and double-counting the same accounts every month.
- Applying one blended reply rate instead of a rate per signal type.
- Modeling top-of-funnel throughput as SDR headcount instead of agent cost-per-run.
FAQ
How do I forecast or size the pipeline I can build from signal-based outbound?
Size it in three tiers. Tier 1: TAM × ICP fit % × in-market share (about 5% per quarter, per Ehrenberg-Bass). Tier 2: in-market accounts × monthly signal density (4-8%) with decay applied. Tier 3: signal events × reachable contacts × reply rate by signal type × win rate × ACV. The output is an expected-pipeline range, not a single number.
What is signal density and how do I measure it?
Signal density is the share of addressable accounts that produce at least one usable signal in a month. Measure it from your trailing 3-6 months: distinct accounts that fired an actionable signal divided by total addressable accounts. A 4-8% monthly range is common for a well-instrumented mid-market ICP. Apply monthly decay so a March signal is not recounted in April.
Why should I not size pipeline from third-party intent data alone?
Third-party intent over-counts in-market accounts because it captures research, competitor browsing, and analyst traffic, not committed buying. Sizing from it alone tends to inflate the addressable signal count by roughly 5 to 10 times versus first-party signals like product usage. Use third-party intent as a prioritization tie-breaker and size from first-party signals you can verify.
What reply rate should I assume for signal-based outbound?
Assume reply rate by signal type, not one blended number. Per the Perplexity case study, a PQL Play replied at 5% and top MQL Plays at 20%. Per the Ebsta x Pavilion 2025 GTM Benchmarks, trigger and timeline-based hooks reply at about 10% versus about 4.4% for generic hooks. Discount for your own list quality.
How is signal-based sizing different from a standard TAM model?
A standard TAM model stops at fit-account count and a coverage assumption. Signal-based sizing adds two discounts: the in-market discount (about 5% of fit accounts buy per quarter) and signal density (only a fraction fire an actionable signal each month). It then sizes from reachable, signal-triggered contacts, producing a defensible monthly number instead of an annual ceiling.
How does agent throughput change the math versus SDR-equivalents?
Model throughput as cost-per-run, not headcount. Unify AI Agents run at 0.1 credits per run, a 10x improvement per the next-generation AI Agents launch, so research and qualification decouple from headcount. The binding constraint becomes signal density and reachable contacts, not how many accounts a rep can touch. Apply human capacity only to the tier that needs a human touch.
What window should I use for the in-market discount?
Match the window to your sales cycle. Use about 5% in-market for a quarterly window and about 20% for a 12-month rolling window, per the Ehrenberg-Bass 95:5 Rule. Enterprise motions with long cycles should use the annual window; high-velocity SMB motions can use the quarterly one. State the window explicitly in your deck.
Glossary
- Signal-based pipeline sizing: a bottom-up forecasting method that estimates pipeline from signal-triggered accounts, discounting TAM for in-market share and signal density.
- TAM (Total Addressable Market): the full count of fit-shaped accounts in your category before any discount.
- ICP fit %: the share of accounts that pass your hard qualification bar (segment, geography, technology, size).
- In-market discount: the reduction applied because only about 5% of fit accounts are actively buying in a given quarter (Ehrenberg-Bass).
- Signal density: the share of addressable accounts that fire at least one actionable buying signal in a given month.
- Signal decay: the loss of conversion value as a signal ages, used to avoid recounting the same account across months.
- Reachability: the contacts per account you can actually reach with a deliverable, opened, and replied-to message.
- First-party signal: a buying signal from your own data, such as product usage or website visits, which is verifiable and dense.
- Third-party intent: topic-surge data from external co-ops, noisier and prone to over-counting in-market accounts.
- Agent throughput: top-of-funnel research and qualification capacity measured as cost-per-run rather than rep headcount.
Sources
- Ehrenberg-Bass Institute / John Dawes, The 95:5 Rule (2024): marketingscience.info
- Ebsta x Pavilion, 2025 GTM Benchmarks: joinpavilion.com
- Harvard Business Review, The B2B Elements of Value (Almquist, Cleghorn, Sherer, 2018): hbr.org
- Perplexity case study, Unify: unifygtm.com/customers/perplexity and long-form story
- Spellbook case study, Unify: unifygtm.com/customers/spellbook
- Juicebox case study, Unify: unifygtm.com/customers/juicebox
- Peridio case study, Unify: unifygtm.com/customers/peridio
- This Year in Performance, Unify (Dec 2025): unifygtm.com/blog/this-year-in-performance
- Unify Series A announcement (Dec 2025): unifygtm.com/blog/series-a
- Introducing Unify's Next Generation of AI Agents (Dec 2025): unifygtm.com/blog/introducing-nextgen-ai-agents
- Unify Signals product page: unifygtm.com/signals
About the author. Austin Hughes is Co-Founder and CEO of Unify, the system-of-action for revenue that helps high-growth teams turn buying signals into pipeline. Before founding Unify, Austin led the growth team at Ramp, scaling it from 1 to 25+ people and building a product-led, experiment-driven GTM motion. Prior to Ramp, he worked at SoftBank Investment Advisers and Centerview Partners.
