The PMF Evidence Gap: what African investors now need to see before a seed or Series A

Executive summary

African startup funding recovered in 2025. Africa: The Big Deal counts $3.2B raised, up 40% on 2024¹. The recovery did not reach more companies. Almost 500 ventures raised $100k or more, flat on 2024, while the number raising $10M or more rose 73% to 69¹. Partech's data shows the same pattern at the early stages: the number of seed deals is down 38% from the 2021 peak, and Series A rounds fell 2% while the average ticket rose 21% to $7.0M².

Fewer, bigger cheques change what a founder has to bring into the room. When an investor writes one cheque in a category instead of three, a claim of product-market fit (PMF) does not carry a round. Evidence does: customers who stay, revenue that repeats, and growth that arrives without being bought.

This paper sets out why PMF claims fail, the evidence investors trust, how VeloDesk PMF Validation weights it, why the L2 Operating Model decides whether it holds up, and a 30-day plan founders can run themselves.

Key findings

• Growth in African startup funding in 2025 came from bigger cheques to fewer companies, not from more companies being funded^1,2.
• Nigerian startup equity fell 21% to $412M in 2025², and Nigeria's share of disclosed African funding fell to 8%, its lowest since 2019³.
• 43% of failed VC-backed startups cite poor product-market fit, and two-thirds of those were early-stage companies that never found a market (global data)⁴.
• No seed cohort since Q3 2021 has reached a 30% Series A graduation rate within two years (US-heavy data)⁵.
• In the VeloDesk PMF Score, the four evidence dimensions (customer evidence, retention, revenue quality, organic growth) carry 5.0 of the 10.6 total weight: 47% of the score.
• The L2 Operating Model is the primary constraint in 91% of enterprises assessed by the African Enterprise Health Index (Crelligent AEHI data; mid-market enterprises, not startups).

1. Capital came back to fewer companies

The headline numbers look like a recovery. Africa: The Big Deal, which counts deals of $100k or more across equity, debt and grants, records $3.2B for 2025, up 40% on 2024¹. That is still about a third below the 2022 peak of $4.8B⁶. Partech, which counts equity and debt separately, records $4.1B, up 25%². The trackers count different things, so we quote one per number.

Look under the totals and the shape of the market has changed:

• The number of funded companies is flat. Almost 500 ventures raised $100k or more in 2025, up just 2% on 2024 and level with 2023. Over the same year, 69 ventures raised $10M or more, up 73%¹.
• Large deals take most of the money. Briter finds that fewer than 5% of deals exceeded $50M, yet those deals took half of all disclosed funding³.
• Debt is doing more of the work. Of Partech's $4.1B, equity was $2.4B (up 8%) and debt was $1.6B (up 63%). Debt is now 41% of capital invested, up from 17% in 2019².

The early stages show the squeeze most clearly.

Source: Partech, 2025 Africa Tech Venture Capital Report².

Nigeria has had the hardest year. Nigerian startup equity fell 21% to $412M, and deal count fell 19% to 83². Nigeria still recorded the most deals of any African market, but its 8% share of disclosed funding was its lowest since 2019³.

For a founder, the meaning is simple. There are fewer rounds to win, and each one is bigger. An investor writing one of 311 seed cheques across a continent, or a $7M Series A, needs evidence that this is the company in its category that will work.

2. Why PMF claims fail

CB Insights studied 431 VC-backed startups that shut down since 2023, 385 of them with known causes. Running out of capital was cited by 70%, poor product-market fit by 43%, bad timing by 29% and unsustainable unit economics by 19%. Respondents could cite more than one reason⁴. CB Insights is clear about how to read the top line:

Running out of capital is "the final cause of death, not the root problem."⁴

Two-thirds of the PMF failures were early-stage companies "that never found a market"⁴. The median time from a startup's last raise to shutdown was 22 months, and about 50,000 VC-backed startups have not raised since the start of 2023⁴. These are global figures, but the mechanism travels: a company raises on a story and runs out of money before the evidence arrives.

The second failure is scaling before the evidence is in. Startup Genome's study of more than 3,200 high-growth startups found that 70% scaled prematurely. None of the premature scalers passed 100,000 users, and their teams were about three times larger than those of consistent peers at the same stage⁷. The study is from 2011 and global, so treat it as direction, not a current benchmark.

PMF claims tend to fail in four ways:

1. Revenue without retention. New sales hide early customers quietly leaving.
2. Growth that is bought. Customer numbers rise with discounts or paid acquisition, and stall when the spend stops.
3. Anecdote instead of cohorts. Strong stories from a few customers, but no table showing what happens to each month's customers over time.
4. Delivery that depends on the founder. Customers stay because a founder personally fixes their problems. That does not survive scale, and investors know it.

3. The evidence investors trust

Investors cannot verify a claim of PMF in a meeting. They can verify data. Three kinds of evidence carry the most weight, because they are the hardest to fake.

Retention

Retention is the signal investors and operators read first. A widely used benchmark, built from a survey of about 20 growth practitioners, puts six-month user retention at these levels⁸:

These are global benchmarks. We found no Africa-specific retention benchmark, so use them as a reference point, not a pass mark.

Revenue quality

Investors look at how much revenue repeats, how much depends on one or two customers, and how much of what was invoiced was collected. A large one-off contract is not evidence.

Organic growth

Customers who refer others without being paid to are the clearest sign that the product works, and they make each new customer cheaper to win.

Time

Evidence takes time to build, and the window is getting longer. Carta's data on 12,249 seed rounds shows no seed cohort since Q3 2021 reaching a 30% Series A graduation rate within two years⁵. The median time from seed to Series A is now 2.1 years, up from 1.5 years in 2019, and the 75th percentile is 3.4 years⁹. Both datasets are US-heavy; we found no equivalent for Africa. Set against a median of 22 months from last raise to shutdown⁴, the conclusion is uncomfortable: many founders will need their evidence before their runway says they need it.

4. How VeloDesk scores PMF

VeloDesk is Crelligent's PMF Validation Platform. It scores product-market fit across 10 dimensions, each scored 0 to 10, and weights them by how much an investor relies on each one.

The VeloDesk PMF Score runs from 0 to 100. It is the weighted sum of the 10 dimension scores, divided by the maximum possible weighted sum (10.6 total weight × 10), multiplied by 100.

The weights are a statement about evidence. Retention carries the highest weight, 1.4×, because it is the signal investors find hardest to fake. Customer evidence and organic growth follow at 1.3×. Together with revenue quality, the four evidence dimensions carry 5.0 of the 10.6 total weight, or 47% of the score.

The arithmetic matters. One extra point on retention adds about 1.3 points to the VeloDesk PMF Score; one extra point on market timing adds about 0.75. The fastest way to raise the score is to improve, and document, the evidence dimensions.

Each VeloDesk PMF Validation produces a PMF Report, an Investor Readiness Assessment showing which dimensions are pitch-ready, and a PMF Narrative Framework for investor conversations.

5. The L2 link: why evidence leaks

Retention and referrals are outcomes. They depend on the company delivering the same value, the same way, to every customer. That is the job of the operating model: the processes, roles and decision rights a team runs on. In Crelligent's ESRE OS, it is L2, the Operating Model layer.

Crelligent's African Enterprise Health Index (AEHI) assesses Nigerian mid-market enterprises across the five OS layers of the ESRE OS. In the current data, the average ESRE OS Performance Score is 44 out of 100, and the L2 Operating Model is the primary constraint in 91% of the 430 enterprises assessed across 9 sectors (Crelligent AEHI data). Because L2 is so often the layer that holds a company back, it carries a 1.4× weight in the ESRE OS Performance Score, the highest of the five OS layers.

The AEHI sample is established mid-market enterprises, not startups. It shows what happens when companies grow without an operating model: if L2 is missing at 400 staff, it was missing at 10.

For a startup, the link to PMF evidence is direct. Without L2, customers get different experiences, some leave, and the ones who stay have less reason to recommend you. In VeloDesk terms, an absent L2 suppresses D3 (Retention & Engagement) and D5 (Organic Growth), two of the three highest-weighted dimensions in the score. This is why D10, OS Readiness, sits in the score at all, and why ESRE OS Lite installs L2 first.

6. A 30-day evidence plan you can run yourself

The plan below uses data most startups already hold. It will not replace a VeloDesk PMF Validation, but it will tell you where you stand.

Week 1: Build the cohort table (D2, D3)

• Define, in one sentence, what counts as an "active" customer. Keep that definition throughout.
• List every customer since launch. Group them by the month they first paid or became active.
• For each monthly group, record the share still active one, three and six months later.
• Compare your six-month figure with the benchmarks above.

Week 2: Test revenue quality (D4, D8)

• Split each month's revenue into repeating and one-off.
• Calculate the share of revenue from your largest customer and your largest three.
• Compare invoiced revenue with collected revenue, month by month.
• Write down what it costs to win a customer and what that customer pays you over time.

Week 3: Trace where growth comes from (D5, D2)

• Tag every customer from the last six months by source: referral, inbound, outbound or paid.
• Speak to customers who stayed and customers who left. Ask what job they hired the product to do and why they stayed or left. Take notes in their words.

Week 4: Check the operating model and write the case (D10, L2)

• Write down who owns onboarding, support, renewals and collections. Mark every step that only works when a founder does it.
• Map each reason for leaving from week 3 to a product gap or a delivery gap.
• Score yourself honestly from 0 to 10 on each of the 10 VeloDesk dimensions, and calculate your weighted score.
• Assemble a one-page evidence pack: the cohort table, the revenue split, the acquisition source mix, the top reasons for leaving, and what you are changing.

After 30 days you will know which dimensions you can lead with, which need work, and whether the problem is the product or the operating model behind it.

7. What this means for your raise

Start building evidence before the raise forces you to. Lead with retention, the evidence investors trust most and founders most often leave out. Show the system behind the numbers, because investors are asking whether the team can keep customers when there are ten times as many. And install L2 before you hire ahead of it: people added to a company without an operating model add cost faster than evidence.

Work with Crelligent Foundry

Crelligent Foundry runs VeloDesk PMF Validation and installs ESRE OS Lite (L1 Business Design, L2 Operating Model and L3 Technology & Platform, with L2 first) for founders preparing a raise. No equity, no cohort. Apply to Foundry or request a quote: it takes about 5 minutes, and we reply with next steps after we review your application.