One of the most common patterns NarrEx surfaces across investment materials is what we call the Ghost Upsell.
The narrative leads with expansion revenue. The investment materials describe a land-and-expand motion. The executive summary emphasises that 30% of projected growth will come from existing customers expanding into additional modules, seats or geographies. The story is coherent. The logic is plausible. The number is specific enough to feel grounded.
Then you open the financial model.
There is no expansion revenue line. The ARR build shows new ARR and churned ARR. There is no cohort-level expansion assumption. There is no module adoption schedule. The revenue projections are driven entirely by new customer acquisition at a flat ARPU — which itself contradicts the upmarket positioning described elsewhere in the investment materials.
The Ghost Upsell is not fraud. It is not even necessarily deliberate misdirection.
What makes it a ghost
In most cases, the expansion narrative reflects genuine commercial intent — the team believes they will sell additional modules to existing customers. The problem is that this belief has not been operationalised in the model. The assumption exists in the story but not in the numbers.
This matters for two reasons.
First, expansion revenue has different economics from new logo revenue. It typically has lower CAC, faster payback and higher margin contribution. If the model is projecting growth purely from new logos while the investment materials describe a business that scales through customer expansion, the underlying unit economics are materially different from what is being presented.
Second, expansion revenue requires specific product, commercial and operational infrastructure to generate. It depends on product depth, customer success capacity and a defined cross-sell or upsell motion. A model that shows no expansion logic provides no evidence that this infrastructure is planned, budgeted or believed to be necessary.
How NarrEx detects it
NarrEx maps each narrative claim to the corresponding model structure. When the investment materials claim that expansion revenue will represent 30% of net new ARR, the system looks for an expansion ARR row, a cohort-level expansion assumption, or a module adoption schedule in the financial model.
If none exists, the claim is flagged as Contradicted — not because the number is wrong, but because the model contains no driver that could produce it.
The evidence trace shows: claim states 30% expansion contribution → model contains no expansion ARR line → no cohort-level driver identified → verdict: Contradicted.
Why it matters for IC discussion
The Ghost Upsell is worth raising explicitly in investment committee discussion because it forces a specific question: is the expansion assumption a deliberate choice to be conservative in the model, or does it reflect the fact that the expansion motion has not yet been designed?
If the former, the model is understating the upside case and the investment materials are describing something real that simply has not been modelled yet. If the latter, the expansion narrative is outrunning the product and commercial reality.
NarrEx does not answer that question. It surfaces the gap so the IC can ask it.
Where expansion hides when it is not in the model
Teams sometimes believe expansion is “in” the model because net revenue retention is above 100% or because churn is low. Those outcomes can coexist with a build that still lacks explicit expansion drivers. NRR is a summary statistic; it does not tell you whether the next dollar of growth is modelled as attach, seat growth, price uplift, or new logos wearing an expansion label in the deck.
Another hiding place is the footnote: “conservative assumptions on upsell.” Conservatism is not a substitute for structure. If expansion is truly conservative, you should still see a placeholder row with explicit sensitivity, or a memo explaining which expansion motions are excluded and why underwriting accepts that exclusion.
A third hiding place is partner or channel revenue that behaves like expansion in the story but like reselling in the model. If partners are expected to drive attach, the model should show partner-sourced ARR or equivalent, not only a faster new-logo ramp with unexplained productivity.
What a reconciled story looks like
The clean version is almost disappointingly literal: the deck’s expansion percentage maps to a module adoption curve or cohort expansion rate; customer success headcount ties to attach timing; gross margin reflects the cost to deliver the modules being sold. You can disagree with the assumptions and still underwrite honestly.
The messy-but-honest version says: we are not modelling expansion in the base case yet; here is the milestone that triggers a model update; here is the downside if attach is slower than hoped. Investors can price milestones. They struggle to price silent optimism.
Post-close: how Ghost Upsells age
If expansion was unmodelled at investment, board meetings become the place where narrative and forecast either converge or diverge again. Track whether attach rates, module revenue lines, and CS ratios move in the direction the story promised. If quarters pass with no structural change in the forecast while language intensifies, you are watching a Ghost Upsell persist into the operating phase.
That is not always fatal. Some businesses genuinely discover expansion motion after product maturity. The portfolio question is whether follow-on capital rewards the discovery only after evidence exists, rather than in anticipation of a motion that still lives primarily in slides.
Sector notes: where Ghost Upsells concentrate
Vertical SaaS businesses often pitch module depth early because industry workflows genuinely layer. The model risk is that module attach is hand-waved while revenue is still mostly new logos in a land-grab phase. Healthcare, construction, and logistics subsectors repeat this shape: rich roadmap slides, thin attach math.
Horizontal collaboration tools pitch seat expansion and workspace growth. The model risk is conflating viral usage with contracted expansion. Usage can lead revenue; it is not the same as modelled expansion ARR until price and packaging catch up.
Infrastructure and developer tooling sometimes describe expansion as deeper consumption of APIs or compute. The model should show unit economics that move with consumption, not only a smooth ARR ramp that could be driven by new accounts alone.
A diligence afternoon: step-by-step
Start by highlighting every sentence in the deck that implies revenue from existing customers beyond simple renewal. Colour-code them. Then open the ARR bridge and highlight every line that could represent those sentences. If the second set is empty, you have located the Ghost Upsell mechanically.
Second pass: search the model for words like “expansion,” “upsell,” “attach,” “module,” and “NRR driver.” Absence of hits is not dispositive—labels vary—but it focuses questions. Ask the CFO what term they would use if expansion were modelled.
Third pass: in interview style, ask customer success how they are measured. If CS is bonused on adoption milestones while finance does not model adoption timing, you have organisational asynchrony that often creates Ghost Upsells even when everyone is competent.
Fourth pass: stress-test. Cut the narrated expansion contribution in half and ask what happens to growth, margin, and hiring. If the answer is “nothing changes in the model,” expansion was never in the forecast to begin with.
Founder-facing language that lands
Founders hear “your model is wrong” as an attack. They hear “help us align the forecast with the story you are telling customers” as partnership. Same technical content, different vector. Ghost Upsell conversations land better when investors frame the gap as forecast hygiene, not a moral judgement.
Offer a concrete deliverable: “We need either an expansion line with assumptions or a deck revision that matches the new-logo-only forecast.” Clear choices reduce thrash. Open-ended scepticism invites deck reshuffling without structural improvement.
When the Ghost Upsell is actually conservative modelling
Sometimes teams know expansion exists but choose a new-logo-only base to avoid overpromising. That can be admirable. The investor-facing requirement is symmetry: the deck should not headline a motion the base case excludes without a clear label. Conservative models paired with aggressive decks are still misaligned artefacts.
If conservatism is the answer, ask for a sensitivity case: base new logos, upside expansion case with attach curve, and explicit probability weights if the firm uses them. You are not demanding false precision; you are demanding visible optionality instead of invisible optimism.
Longer horizons: expansion and moat claims
Ghost Upsells interact with moat language. “Sticky platform” and “land-and-expand” often arrive as a pair. If expansion is unmodelled, the moat claim may still be directionally true while being unpriced. Decide whether you are underwriting today's forecast or tomorrow's strategic thesis; do not accidentally do both without noticing.
Boards can help by asking for a single chart each quarter: expansion-attributed revenue versus plan, with definitions held constant. The chart is often boring when things work and loud when they do not. Boring charts are an underrated governance luxury.
We will stop here, having walked from definition to detection, hiding places, reconciliation, portfolio ageing, sector texture, a practical sequence, founder tone, conservative-modelling nuance, and governance habits. The Ghost Upsell will keep appearing because expansion is how software likes to tell stories about itself. Your job is to ensure the spreadsheet is invited to the same conversation.
One-page IC appendix you can reuse
Title it “Expansion evidence map.” Row A: quoted claim from the deck. Row B: model line or explicit statement that the claim is excluded from base case. Row C: artefact requested to close the gap (module curve, CS plan, cohort cut). Row D: owner and date. Most committees spend less time arguing when the blank spaces are visible on one page.
If Row B says “excluded,” Row E should record valuation implication: are you pricing new-logo economics while reading expansion prose? If yes, note the delta explicitly so reserves and follow-on rules stay honest.
Reuse the appendix across companies. Habit beats heroics. The Ghost Upsell dies in firms where mapping claims to drivers is as routine as checking cash balance.
Reading sell-side materials without inheriting their optimism
Banker models often inherit deck language because speed matters in a process. Treat third-party materials as a second narrative layer, not as independent confirmation. Ask the company directly whether the sell-side expansion assumptions match internal forecasting. Mismatches between banker deck and management model are more common than investors like to admit.
When you cannot get a clean internal model, require a management bridge from their operational forecast to the external summary. Absence of that bridge is a process risk separate from the Ghost Upsell itself, but the two risks stack unpleasantly.
Syndicate dynamics pressure everyone to move quickly. The counterweight is a single written question logged before soft circle: “Where is expansion in your model, and if nowhere, why does the headline still say land-and-expand?” The answer belongs in the data room, not only in a verbal sidebar.
Closing the loop
If you remember nothing else: expansion in the pitch without expansion in the model is not a small cosmetic gap. It is a different business with different risk. Name it early, reconcile it honestly, and price what remains uncertain. The Ghost Upsell survives on speed and polish; patient mapping kills it without drama.
Carry the question into diligence dinners, management sessions, and reference calls. Ask customers how they actually buy the second module, how long it took, and who blocked it. Then compare answers to the expansion timing implied in the story. Field truth is not the whole truth, but it is a useful cross-check when spreadsheets stay silent.
That mix of model discipline and customer texture is what turns this from a pattern name into a working habit. Patterns are shortcuts; habits are what keep shortcuts from becoming stereotypes. Use the name to find the gap, then do the slower work of closing it with numbers, owners, and dates.
Read it twice if you are sponsoring the deal: once for speed, once for silence—the places where the model has nothing to say usually matter more than the places where it speaks confidently.
The Ghost Upsell is one of five recurring narrative-model misalignment patterns NarrEx detects systematically. The others are the Scaling Paradox, the Lagging Hire, the Enterprise Illusion, and the Retention Friction.