Your Seed Deck: The Business. The Market. The Investor.

Start with the claim you lean on most

Perhaps an investor has just said no and you are wondering what to do next. Perhaps you have built something you are convinced could change the world. Either way, you are carrying a set of claims about your business, and some of them matter far more than others.

A deck can make the problem vivid, the market enormous and the team credible. It can show a growing user count, a row of company logos, a striking revenue-adjacent number, or a chart that rises neatly from left to right. Each may be useful evidence. None, on its own, tells you what the market has actually done.

That is why other founders' decks are worth your time. A seed deck is a set of choices: what a founder shows tells you which claims they wanted outsiders to see, and you can read your own deck the same way. If a deck names a large market but gives no buyer-level evidence, the reader is left to bridge the gap. If it shows $1 million of transactions without stating revenue retained, the reader must infer the company's economic share. If it mentions retention without a cohort or repeat rate, it is unclear whether customers have demonstrably returned.

The line that matters most is rarely between a strong pitch and a weak one. It is between a persuasive claim and what customers have actually demonstrated.

We read 113 early-stage decks from a public library of fundraising presentations and 88 that founders had posted themselves, a separate group that does not begin with companies described as having raised. A small number of hand-picked decks fill in what broad counts cannot explain. We asked commercial questions, not design ones: What evidence of real buyers, users and payments is visible? Does a figure mean company revenue or something else? Is there evidence of repeat behaviour? How does the company propose to reach its next customers? What does the deck say about the proposed use of funding? How confidently can a reader distinguish achieved results from forecasts?

None of this means an early company needs every answer before it starts. A founder at concept stage may reasonably have problem interviews, a pilot commitment and a hypothesis about pricing rather than live revenue. Being early is not the problem. Treating an untested proposition as settled is.

Before you read on, pick the slide you are proudest of and ask what a stranger would have to assume to believe it.

What we found, and why it matters to your business

Every count here describes what a public copy of a deck disclosed, which is rarely everything a team knew. Unless another group is named, the findings describe the decks in Figure 1.

1. Money and customer signals were common, but interpretable revenue was not. Reaching one of the two highest evidence categories took either a stated monetary figure or some customer or revenue signal, which does not mean a deck demonstrated paying customers. Beyond the decks that labelled a figure as their own revenue, four more amounts were ambiguous.
2. The big number was not always money earned. Thirty-seven decks showed a monetary figure. Of these, 13 primarily presented other or unclear measures, such as transaction value, contract or pilot value, annualised sales, or an inadequately labelled amount, and another four were borderline revenue claims. The label, unit, period and economic meaning matter.
3. Customer presence was often easier to see than customer value. Twenty-six of 113 (23%) presented customer or revenue-related signals without a usable figure. Logos, unnamed customer counts, upward-moving but unnumbered charts and redacted public slides can all communicate traction while leaving a reader unable to evaluate its scale.
4. Repeat behaviour and acquisition economics were thinly documented. Fewer than half of the decks that mentioned repeat behaviour gave an observed rate, and just over half of those that described a route to future customers made it concrete.
5. Pitch format and disclosure moved together. Thirty of 63 demo-day decks (48%) showed a monetary figure, compared with seven of 50 other investor decks (14%). Those populations also differ in year, selection and public-redaction patterns, so read this as a pattern in disclosure rather than the effect of a template.
6. A separate founder-posted sample shows what decks outside a funding library disclosed. In 40 of its 88 decks, the public copy showed no customer evidence.
7. The clearest founder lesson is to connect the evidence. A customer logo should connect to the relationship it represents. A financial figure should connect to the revenue actually retained. A growth chart should connect to a period and denominator. A go-to-market plan should connect to a specific reachable buyer and an observable next test.

Which decks these numbers describe

This report rests on 201 decks in two separately analysed groups, 113 funded-library decks and 88 founder-posted decks, whose results are never pooled into one rate. Both sit inside a wider evidence register of 256 entries representing 255 distinct ventures, because Dimension and Trash Warrior are two listings for the same venture. The register is an inventory rather than one sample: it includes records left out of the headline analysis, pilot material and deliberately selected examples. Appendix A sets out how each group was drawn and what was excluded.

1.1 The funded-library analysis: 113 decks

The first group came from a public pitch-deck library. The library labels these presentations as associated with fundraising. That label does not confirm that any particular round closed or that the public deck was exactly what investors saw. Within the 113, 63 were accelerator demo-day presentations and 50 were other investor decks, published at different times and in different ways.

1.2 Founder-posted decks: 88 analysable presentations

The second group came from a separate collection of seed and pre-seed presentations that founders had made public themselves. It looks different: it includes small, local, student and one-person ventures, not only companies that appeared in a funding library. If you are early and self-funded, these decks may look more like yours.

1.3 Deliberately selected case presentations

A smaller set of historical decks was read as illustrations, chosen on purpose to show particular issues: the difference between user counts and revenue, acquisition costs beside sales, retention, the meaning of soft commitments and what a fundraising story can and cannot settle. None of them enters a percentage in this report.

1.4 What the evidence levels mean

Each deck was assigned the highest level of evidence visible in its public copy. These levels describe the kind of evidence disclosed, not a grade of venture quality or a probability of investment success.

A genuine early-stage experiment at level D can mean a great deal. A large level-A transaction figure can reveal less than a small but clearly labelled recurring-revenue number. Find where your own strongest evidence sits, and you know what the next level up would take.

What 113 funded-library decks showed about their customers

2.1 The evidence distribution

There are two ways to read this chart. First, 63 of the 113 decks (56%) showed something at level A or B: a monetary or customer-related signal. Second, 50 (44%) did not reach either of those two categories in their public presentation.

The 63 are not all proof of paying customers: level A may contain marketplace transaction values or other money-like measures, and level B may contain customer logos without a payment statement.

The market question is more exact: What relationship between the company and a real customer does the slide actually establish? A usage chart establishes that something happened in the product. A named paying customer and the amount paid establish something different.

2.2 “We have customers” can mean several things

A deck might show the names of companies, a count of registered organisations, interviews with prospective buyers, a pilot arrangement or actual payment. These signals are not equivalent. A logo could reflect a customer, a design partner, a past employer, an integration or a prospect; without a legend, a reader cannot know.

In the public deck for Blurbiz, client logos and a price list appear on separate slides (slides 2 and 6). The deck does not itself connect the logos to an identifiable billed amount. In Waterplan, a claim that ten Fortune 500 companies used the product (slide 4) does not by itself state what they paid or identify the relationship. LoanWell presents an annual price range and 13 customer logos (slides 11 and 12); the logos alone do not show 13 paying customers.

The issue is not dishonesty. These are different ways of indicating progress. The gap is between showing that a relationship may exist and demonstrating its commercial value, and that gap is where an investor's first reasonable question sits.

Storyboard presents a rising chart of paid seats but lacks the numerical scale needed to calculate what the movement represents (slide 5). A graph without values can show direction while concealing magnitude. VideoPeel is a reminder of another limitation: its public copy replaces a monthly recurring-revenue number with a placeholder (slide 14). A redacted public deck does not show what investors may have seen privately.

A large money figure is not necessarily company revenue

Of the 37 level-A decks containing a monetary figure, 20 explicitly labelled it as the company's own revenue. Four others are borderline: Connected2me labels a figure as profit, Pepper Bio shows a range or ceiling rather than an achieved revenue result, and Catnip and Voxeet show amounts whose measure is not clear from the slides. Thirteen more represent other or unclear measures.

“37 decks showed revenue” would therefore be a misleading headline.

3.1 Transaction volume versus money retained

Homigo presented a reported $780,000 run rate in transaction value (slide 2). The deck charts continue to January 2017 despite an earlier library-assigned year. A rental or marketplace platform may facilitate a large volume of payments while retaining only a relatively small portion, so transaction value is not necessarily Homigo's income.

By contrast, the iGrow deck places a 20% gross-profit claim beside $1.2 million of transaction value (slide 8). That additional ratio helps a reader ask how much economic benefit could accrue to the company. That makes it more interpretable than transaction value alone.

If you run a marketplace, four questions sit behind a transaction-volume headline: What was transacted? Over what period? What fees did the business retain? What costs were incurred to generate and serve those transactions? You can pull all four answers from your own records.

3.2 Contracts, pilots and annualised numbers

Butlr reported $1.07 million of total contract value signed (slide 4). A signed contract is evidence of a commercial relationship, but total contract value is not automatically equivalent to cash collected or revenue earned in the period. Tallyfy described $600,000 in pilots (slide 3), while the public slide as recorded did not resolve whether this was contracted, billed or prospective value.

Shippo used a $1 million run-rate figure adjacent to a chart with much smaller axis values (slide 3). An annualised run rate can be a meaningful operating indicator, but the underlying time interval, actual recorded revenue and method of extrapolation need to be stated. For an early business, the difference between one strong month and a recurring twelve-month pattern is substantial.

Bill Trim charted “gross revenue” without clearly demonstrating the amount the company kept from a service that generated savings for customers (slide 7). A label can be technically plausible yet insufficient to establish how the commercial model works.

The principle for your own deck: name the measure before using it as proof. “Our customers processed £500,000 through the platform” communicates something different from “we earned £50,000 of revenue,” and both differ from “we have signed agreements worth £500,000 over two years.”

3.3 What clearer financial evidence looks like

The recorded Infraspeak presentation puts revenue, monthly recurring revenue and churn together on slide 11. That makes the commercial argument easier to examine: there is money, a recurring component and an indication of customer loss.

We also encountered a library-produced redrawn summary of mcSquares that places order value, acquisition cost, margin and lifetime value together. Because we did not read the original founder presentation behind that summary, we treat it as an example of a presentation format rather than primary evidence of the company's financial performance.

Mero Technologies presented growth in recurring revenue alongside buildings and distributor counts (slide 12). The strength of the presentation is the connection between an economic outcome and the operational mechanism behind it, rather than a revenue headline floating on its own.

Retention: the question after the first customer

Winning an initial customer is one test. Discovering whether that customer returns, renews, expands or continues to gain enough value to pay is another. A startup that has demonstrated willingness to buy once may still have an unresolved question about durable demand.

In the 113-deck analysis, 20 decks (18%) mentioned retention, repeat use or lifetime value, but only nine (8%) provided an observed retention or repeat rate. Six more supplied lifetime-value estimates; five mentioned retention without a usable figure. These categories should not be collapsed into “20 decks proved retention.”

A retention percentage is not automatically enough. It matters which customers were counted, when they joined, what constitutes staying, how long they were observed and whether the measure concerns users, accounts, revenue or another unit. A lifetime-value estimate can also be useful but depends on assumptions about future behaviour and margins.

Homigo stated a 24-month customer lifetime value while the visible historical chart covered substantially less time (slide 3). Such a figure may be a modelled estimate rather than 24 months of observed retention.

You do not need a retention slide before you have the operating history for one. You need a next experiment that makes repeat behaviour measurable. A services founder may look for a second purchase or expanded engagement. A subscription founder may track cohort renewals. A marketplace may examine whether both sides complete another transaction without extraordinary incentives.

For a pre-revenue team, the honest statement may be that retention is not yet measurable. That is more useful than projecting a confident lifetime value from a few weeks of data: it says what you know now, and how you plan to find out the rest.

How will the next customers actually arrive?

A pitch often includes familiar acquisition vocabulary: partnerships, paid media, community, referrals, outbound sales or a distributor network. These are channels. They are not necessarily a tested mechanism for reaching a specific buyer at an acceptable cost.

The analysis found 27/113 decks (24%) with some description of a route to the next customers, but only 15/113 (13%) with a concrete route under the stricter definition. The additional twelve named something relevant without enough mechanism or quantity to establish how the next sale would likely happen.

A concrete route can take different forms: a defined customer type and trigger, an identifiable way to reach them, a tested conversion process, a measured sales cycle, a known distributor or a credible pipeline with clear stages. It need not be a sophisticated growth model. It needs to be specific enough that a founder could test it.

ProcureNow showed an average sales cycle of about three months next to its pipeline (slides 8 and 9). That connects a proposed sales route to a time requirement. Laoshi described tutor acquisition cost and the students a tutor brought onto the platform (slide 14), drawing attention to a mechanism of distribution rather than a generic channel label.

For a B2C founder, this often becomes an access problem before it becomes a marketing-budget problem. Which group of people experiences the problem frequently? What event makes them look for a solution? Where can they be reached now? What evidence would show that the initial acquisition method can work repeatedly? A market-size slide rarely answers those questions.

A useful next-customer statement therefore has a testable shape: We believe people in this situation can be reached through this route, will respond to this proposition, and will take this observable commercial action. The next experiment tests the weakest link in that sentence.

A big market and a strong team are not yet a market response

Team slides were present in 99 of the 113 decks (88%), and market-size claims in 83 of the 113 (73%). Among the 83 decks that sized the market, 61 used a top-down total without a demonstrated path from the large category to a reachable first segment.

Sizing the market still matters. A company needs to know whether a commercial opportunity could become large. It just cannot stand in for proof that customers will recognise the problem, choose the offer and pay.

A familiar example is the jump from “there are millions of potential users” to “we will capture one percent.” The first figure describes a possible population; the second is a forecast about winning customers. Without a reason for selecting those buyers, an access route and evidence of willingness to engage, the conversion between the two is not established.

Team experience also matters, particularly where the company is pre-product. Kamino described founders with technology and fintech experience, including one documented earlier exit, while the product was still to be built (slides 2 and 5). The deck demonstrates how a team can present expertise before presenting customer results, not that expertise removes the need for later market learning.

Revolut provides another instructive early-stage example. Its 2015 public deck described 4,800 sign-ups accumulated in one month and nine deals, including one with MasterCard (slides 9 and 10), in the context of a £1.5 million funding request. Those are forms of pre-launch interest and partnership development, not a demonstration of already recurring customer revenue. They help explain why early teams can reasonably present evidence appropriate to their stage.

The lesson for any founder: use the strongest evidence that exists, and let customers, rather than somebody else's credibility or fundraising context, show that the demand is there.

Format and what founders disclose

Demo-day decks and other investor decks differed markedly in which evidence level appeared on the public slides.

A structured demo-day presentation tends to require a traction moment. A general investor presentation may devote more space to the team, market and narrative while leaving detailed evidence for a later conversation. The pattern fits that difference, though format is only one of several things that vary between these decks.

Public versions can also be edited. Six decks had figures or material explicitly removed from the public version, and counting one more company whose revenue was explicitly withheld makes seven.

Copying an accelerator template is not the lesson. A format that forces clarity about traction brings a critical question forward, and you can ask it of your own deck today: What can we point to, in numbers or in words, that is happening in the market?

What the founder-posted decks showed

The second set of decks offers a different perspective. Forty of its 88 decks showed no customer evidence in the public copy. That describes the deck rather than the business behind it: a founder may have had customers the public slides never mentioned.

The research also examined publicly available information about what happened to the ventures afterward. These records were incomplete and do not support a sufficiently reliable comparison of subsequent company outcomes. We therefore draw no survival conclusion from this analysis.

A missing record can mean many things. Some ventures were too small to leave a searchable footprint. Some could have changed names, become parts of other businesses, or continued privately.

If your own deck shows no customer evidence yet, you are far from alone, and the next question is a practical one: What small piece of direct customer evidence could move my business from an attractive proposition towards a commercial fact? That could be a paid pilot, a repeat order, a signed trial that starts on a date, a response from a precisely defined buyer group, or a measured acquisition journey.

What selected decks can teach us

We chose these decks because each exposes a commercial question clearly enough that you may recognise it in your own business. They are illustrations: they cannot establish a survival or fundraising rate. We stay with what the slides show and do not infer a company's later outcome from its deck.

9.1 Flowtab: sales activity is not the same as captured revenue

The recorded Flowtab traction slide reports seven bars, 2,200 users, approximately 120 orders a week, $12,450 of total sales across three months, and a cost of $1,500–$2,500 to win each bar (slide 7).

That combination invites a valuable question. If “total sales” means drink sales going through the product rather than revenue retained by Flowtab, what amount did the company actually earn? How many transactions repeated, and what did it cost to keep each venue active? The slide does not provide enough information to conclude whether acquisition costs were recovered, whether the model could sustain itself or what happened later to the company.

The next useful commercial test for a business like this is to put acquisition cost, revenue retained by the company, the period and repeat behaviour in one view. A large volume of activity may validate interest. It does not automatically settle the economics.

9.2 Manpacks: even an impressive retention label needs a denominator

The public Manpacks deck describes more than 1,800 paid subscribers, 10% monthly growth, a funnel and a measure labelled “65% Q1–Q4 retention” (slides 10 and 12). This is stronger customer-related disclosure than a deck that merely says people love the product. Yet an investor would still reasonably ask exactly what was retained, which cohorts were measured and how the figure was calculated.

Any founder quoting a rate like this should put the cohort and the calculation beside it.

9.3 Bliss.ai: reported paid customers do not explain every commercial unknown

The public Bliss.ai deck reports 70 paying customers, an $84,000 annualised revenue run rate, a margin claim and named enterprise relationships (slides 6 and 8). They show more than a sign-up count, and they leave useful questions open about customer lifetime, repeat purchasing, acquisition economics and how the company captured value.

9.4 Castle: unit counts, price and the missing bridge

Castle showed rental units under management rising from four to 49 since a soft launch on 1 January and stated a price of $79 per unit per month (slides 7 and 8). Those two pieces of information can help a reader understand the possible economics of a property-management offer. The next questions are whether every unit paid the stated fee, whether revenue was collected and whether retention was strong.

The test here is to link the growth of units to realised billing, servicing cost and retention.

9.5 Standard Treasury: the funding ask and the use-of-funds budget are different

Standard Treasury requested $10 million towards UK banking authorisation on slide 13, while the 24-month budget shown on slide 14 was approximately $9 million. The request and the spending plan serve different purposes. Neither should be silently substituted for the other.

Slide 25 contains startup logos and references to soft commitments; the logos do not establish five separately confirmed binding commitments. That is a useful distinction for any founder presenting prospective demand, design partners or letters of interest. A commitment can be important evidence, but the document must say what has actually been promised, by whom and under which conditions.

9.6 Everpix and Moz: reporting an indicator is not answering every business question

The Everpix deck reported 40,000 users and a claim that 13.1% of sign-ups paid, but did not provide a company revenue figure in the cited slides (21 and 25). The Moz deck presented a $10.8 million revenue run rate, a subscriber count and several unit-economic measures (slides 20 and 21). These are strikingly different kinds of deck evidence.

What matters is the evidence question each raises: how do paying share, actual revenue, cost to acquire, renewal and gross margin combine into a model that can be understood and tested?

What an investor can reasonably understand from a deck

These decks are not a recipe for fundraising. Nothing here shows that including a slide, quoting a metric or avoiding a phrase makes investors say yes, and a public deck is only one piece of a wider decision.

What a funding audience can reasonably take from a deck is which links in the commercial chain are established and which are still open. A compelling pitch about a large market raises the next question: can these buyers be reached? A transaction-volume chart asks how much the company captured. A pilot logo asks whether there is a contract. An optimistic lifetime value asks what a real cohort has done. A list of marketing channels asks how the next sale happens.

Investors sometimes finance potential. Founders build evidence. The two are related, and this report is about the second. A strong story can explain why a market matters and why a team might win. Investor interest cannot show whether customers recognise the problem, can be reached, act on the proposition, pay and return, and a founder goes on learning that from the market after any funding conversation, whichever way it went.

That matters more now that building and presenting software can be faster than learning whether a customer genuinely wants it. A founder can produce a credible product demonstration long before discovering whether the buyer feels the problem strongly, has the authority to purchase or will exchange money for the solution. How fast you can make something is not the evidence you need to choose your next step.

Six commercial questions a founder can use immediately

Together these six are the practical core of this research.

Work through all six in the exercise above, with room for your own notes. Continuous reading sets them out here in full.

1. 1Who has demonstrated the problem?

   Can we point to specific buyers or users who independently described a costly, frequent or urgent problem? How were they selected? What would contradict our hypothesis?

2. 2What customer action has actually occurred?

   Was there an interview, waitlist sign-up, pilot agreement, activation, repeat use, payment or renewal? Each action has a different meaning. Which one do we currently have?

3. 3What money did the company earn?

   If we use a financial figure, is it revenue recognised by our company, money passing through the platform, contract value, gross profit, forecast or an annualised run rate? What is the time period, currency and source?

4. 4Do customers repeat the action?

   Are they coming back, renewing, expanding or buying again? If we are too early to measure that, what experiment would create the first observable repeat signal?

5. 5Can we describe how to reach the next buyer?

   What is the customer's trigger, where can we access them, what message earns a response and what is the next commercial action? Can we test this with a modest number of reachable buyers?

6. 6What is still unproven, and what is the smallest useful next test?

   Rather than filling a missing slide with assumptions, identify the single uncertainty most likely to change the decision to build, sell, spend or raise.

None of this is a reason to hesitate. You can have a real business in progress without knowing the full economics.

From business claim to next commercial test

Your deck may have a communication problem, an evidence problem or both. The right next step depends on which.

The route has four steps. It is a discipline rather than a mandatory sequence, and it works at any stage:

• Business claim: What you believe, have built, claim and propose about the buyer, problem, offer or market.
• Market response: What somebody outside the founding team actually said, did, signed, paid or repeated.
• What an investor can reasonably understand: What those two establish for a funding audience, and the most important question they leave open.
• Next useful commercial test: The smallest ethical, affordable action that could strengthen or change the decision.

What to test next depends on what you have already demonstrated:

• Reliable revenue and retention data, labelled poorly: communicate the existing evidence accurately.
• A valuable problem, but no direct buyer observations: the next task is not a better slide. It is an encounter with actual potential customers.
• Enthusiastic interviewees but no purchase: test willingness to commit.
• Pilots but no renewal or repeat behaviour: test repeated value, by looking at retention, value delivered and the conditions of a second purchase.
• Customer evidence but no path to more customers: test access and acquisition economics.
• Paid customers but unknown acquisition cost: the commercial unknown is no longer whether people will pay at all.

This is where GTM Right comes in. Its purpose is to help a founder investigate the commercial reality behind the proposition. Someone who recognises an untested buyer assumption or a missing evidence link can begin with a free Stage 1 assessment and, if appropriate, continue to a deeper commercial-readiness process and a Commercial Readiness Pack.

That journey exists to help you make consequential choices (building, pricing, launching, seeking funding or changing direction) with a clearer view of what the market has shown.

From your strongest claim to what customers have demonstrated

These decks are history. The question they leave behind is about your next decision.

Across them, the absence of clarity was often more revealing than the absence of any customer. There were signs of demand, figures, partnerships and projections. The unanswered questions were about what those meant: Who paid? How much did the company keep? Did they come back? How will the next customers arrive?

You can see your own customer records in a way no one can see these decks, so you can start answering those questions today. You do not need the perfect presentation first. Sometimes the next step is building. Sometimes it is selling. Sometimes it is getting closer to the buyer before spending another month on the product. Sometimes it is changing a claim, a price, a segment or the entire proposition.

The most useful question a pitch can lead you to is not “Does this sound investable?” It is this: What would it take to turn your strongest pitch claim into something customers have actually demonstrated?

Which claim would you least like a customer to test? That is the one to take further. Each pass leaves you with something more solid to build on, and a clearer decision about what to do next.

Methodology

Evidence definitions: six distinctions this report keeps

How the research was done

Scope. This publication reports on 201 public startup decks, analysed in two separate cohorts that are never pooled into one rate: 113 funded-library decks and 88 founder-posted decks. The broader evidence register holds 256 entries representing 255 distinct ventures. It also contains excluded records, pilot decks and 10 selected historical cases, none of which enter a rate.

Study type. Retrospective content analysis of historical public presentations, paired with exploratory public-source company tracing. Not a prospective trial, an investor-decision experiment or a representative census of all startups.

Data capture. First source library captured 5 November 2026; second source collection captured 6 November 2026. Individual library years are not necessarily the dates of the original founder-created slides. For example, the New Era ADR slide dates itself to 2021 despite a 2022 library label, and the mcSquares redraw refers to July 2021 despite a 2020 library label.

Frame and draw. First source: 1,612 listing entries; 778 early-stage labels; 726 year-eligible, carrying years between 2012 and 2023; 121 sampled by recorded stratified draw using the fixed seed 20261105. Three selections (RAPT.FM, We Are Onyx and Well Track) could not be read and were not replaced, although the stated protocol allowed same-year replacement. Five of the 118 readable presentations were excluded as not the relevant kind of investor pitch, including Tinder's 2012 internal concept presentation, leaving 113 analysed. The original draw script was not retained; a reconstruction using the recorded seed reportedly reproduced the recorded positions. Two source aliases, Credit Stacks/Jasper and Well Track/WellTrack, appeared twice in the 726-entry frame, which may have affected sampling probabilities. The original draw has been retained.

Second source. 983 distinct-source entries after overlap filtering against the first library; 412 dated 2012–2019; 125 drawn, of which 30 formed a pilot, kept apart and never counted in the 88, and 95 a planned test; seven test entries excluded as not eligible investor decks; 88 analysed.

Selected cases. Historical illustration only, intentionally chosen, excluded from all headline sample percentages. Their deck content is not evidence that a later adverse company event occurred.

What counted. Each deck was given the highest level visible in its published slides: A, a stated monetary figure; B, a customer or revenue signal without a usable monetary figure; C, users, downloads, usage or engagement without customer or revenue evidence; D, pilots, waitlists, planned trials, pipeline or other commitments; E, no customer evidence in the public deck. The table above sets out how money figures, retention and acquisition routes were then separated.

Figures as presented. Figures were transcribed from public slides and classified according to rules. They are presentation disclosures rather than company accounts.

Research method. AI-assisted methods were used to read and classify the publicly available presentations. The classifications were not independently second-coded. The research reports what published deck material disclosed, with source references and definitions provided.

How far these findings go

Investor decisions stay private. Funded-library decks were not balanced with a systematic sample of rejected rounds, so the research cannot say why an investor invested or declined.

Historical decks. Most date from 2012–2023. They illustrate enduring commercial-evidence distinctions, and they are not a survey of what investors ask for today.

Later company status. The study looked for dated public evidence of continuing operation or acquisition. Public outcome information is uneven: a parked domain, missing website, dissolved subsidiary or retired product may not represent the status of an entire venture, and “acquired” says nothing by itself about investor returns. These records were incomplete, so no comparison of later company outcomes is reported.

Known limits. Single-library dependence, selection bias toward published/funded decks, redacted public copies, time-period differences, duplicated company aliases in the frame, three unreplaced unreadable selections, ambiguous deck years, nonrandom selected cases, incomplete visibility of private diligence and imperfect public tracing of subsequent ventures.

Public dataset policy. Unchecked third-party library funding amounts are not appropriate as a public research field without separate verification. Internal original records are preserved to make corrections traceable. Where a closure, failure or acquisition could not be substantiated sufficiently for publication, this report leaves it out.

Source index for cited decks

The following are the source pages or source hosts carried in the research material. Slide references throughout refer to the publicly available deck as read by the research. These links point to the sources as recorded, and some pages may since have changed, been redrawn or stopped working.

Deck rights stay with their owners. The report cites deck pages and slide references and reproduces no deck images. A public URL is not permission to republish images or entire presentations.

1. S01First study frame Public source for First study frame (opens in a new tab) ↗public early-stage deck library
2. S02Second study frameInvestor Slides public seed/pre-seed collection, as recorded in GTM Right's research register
3. D01Blurbiz Public source for Blurbiz (opens in a new tab) ↗slides 2, 6
4. D02Waterplan Public source for Waterplan (opens in a new tab) ↗slide 4
5. D03LoanWell Public source for LoanWell (opens in a new tab) ↗slides 11, 12
6. D04Storyboard Public source for Storyboard (opens in a new tab) ↗slide 5
7. D05VideoPeel Public source for VideoPeel (opens in a new tab) ↗slide 14
8. D06Homigo Public source for Homigo (opens in a new tab) ↗slides 2, 3; charts extend to January 2017, later than the 2015 library label
9. D07iGrow Public source for iGrow (opens in a new tab) ↗slide 8
10. D08Butlr Public source for Butlr (opens in a new tab) ↗slide 4
11. D09Tallyfy Public source for Tallyfy (opens in a new tab) ↗slide 3
12. D10Shippo Public source for Shippo (opens in a new tab) ↗slide 3
13. D11Bill Trim Public source for Bill Trim (opens in a new tab) ↗slide 7
14. D12Infraspeak Public source for Infraspeak (opens in a new tab) ↗slide 11
15. D13Mero Technologies Public source for Mero Technologies (opens in a new tab) ↗slide 12
16. D14New Era ADR Public source for New Era ADR (opens in a new tab) ↗slide 9, internal date 2021
17. D15ProcureNow (Govlist) Public source for ProcureNow (Govlist) (opens in a new tab) ↗slides 8, 9
18. D16Laoshi Public source for Laoshi (opens in a new tab) ↗slide 14
19. D17Kamino Public source for Kamino (opens in a new tab) ↗slides 2, 5
20. D18Revolut Public source for Revolut (opens in a new tab) ↗slides 9, 10
21. D19mcSquares Public source for mcSquares (opens in a new tab) ↗library-redrawn summary; original source not established
22. D20Mmob Public source for Mmob (opens in a new tab) ↗slide 7
23. D21Flowtab Public source for Flowtab (opens in a new tab) ↗Public angel presentation held in the Alexander Jarvis seed-deck collection; research register references slide 7
24. D22Manpacks Public source for Manpacks (opens in a new tab) ↗Public seed presentation held in the Alexander Jarvis seed-deck collection; research register references slides 10, 12
25. D23Bliss.ai Public source for Bliss.ai (opens in a new tab) ↗Public seed presentation held in the Alexander Jarvis seed-deck collection; research register references slides 6, 8
26. D24Castle Public source for Castle (opens in a new tab) ↗Public angel presentation held in the Alexander Jarvis seed-deck collection; research register references slides 7, 8
27. D25Standard Treasury Public source for Standard Treasury (opens in a new tab) ↗slides 13, 14, 25
28. D26Everpix Public source for Everpix (opens in a new tab) ↗Public Series A slides available through SlideShare; research register references slides 21, 25
29. D27Moz (SEOmoz) Public source for Moz (SEOmoz) (opens in a new tab) ↗Public growth-stage slides available through SlideShare; research register references slides 20, 21

The research's complete internal evidence register retains company aliases, deck URLs, slide evidence and supporting source notes.

Complete company and deck-source index

This index reproduces the 256 entries in the research inventory: 121 funded-library decks, 125 founder-posted decks and 10 selected historical cases. These are 256 entries, not 256 distinct ventures; Dimension and Trash Warrior refer to one company, giving 255 distinct ventures across the inventory. The analysed quantitative samples are smaller (113 funded decks and 88 founder-posted decks).

Each link points to the public source page recorded in the research material; it may be a deck-library page rather than a direct PDF or hosted slide file. Links are historical research references. As of 8 November 2026, two recorded source pages no longer load; they appear below as text rather than links. Three sampled decks (RAPT.FM, We Are Onyx and Well Track) were reported unreadable. Fourteen funded-library entries carry a note on their format, such as a teaser, an overview document or a deck that appears later than seed stage. Thirteen of those remain in the 113; the fourteenth, an internal concept deck, was excluded.

The list is not a ranking, endorsement or company-failure directory, and it carries no company-status data.

Source of the index: GTM Right's research register.