African funding trackers disagree because they count different things

TechCabal Insights reported $1.44 billion of African startup funding in the first half of 2026. Africa: The Big Deal reported close to $1.4 billion excluding exits and counted 190 ventures raising at least $100,000. The proximity of the rounded totals does not make the datasets interchangeable.

Funding trackers build a market from public announcements, direct submissions and research. They decide which companies, instruments, amounts and dates enter. Comparing their headlines without those decisions creates a false discrepancy.

Definitions carry the comparison.

Row one: minimum size

A tracker can include every disclosed round it verifies or set a minimum such as $100,000. The threshold changes company count more than total dollars because excluded transactions tend to be small.

Thresholds can apply to a round, a company's cumulative annual funding or the disclosed portion. A company announcing “six-figure funding” presents another choice: estimate, exclude or count without value. Read the note.

A minimum improves comparability and reduces noise. It also makes pre-seed activity less visible. Analysts studying early capital need a dataset built for that question.

Row two: debt and grants

TechCabal's H1 article splits its $1.441 billion rounded total into $818 million equity, $614 million debt and $9 million grants. The Big Deal described an approximate two-thirds equity and one-third debt mix under its method.

Debt can mean a corporate loan, asset finance, warehouse facility or another disclosed instrument. Trackers may count the committed ceiling or the initial draw. A grant may be non-dilutive company funding or programme support shared among recipients.

Totals need instrument columns. An equity investor cannot use a debt-inclusive aggregate as a measure of available venture equity. A lender may care about committed credit and the number of borrowers instead.

Convertible notes complicate the split. Legal debt that is expected to convert may appear in an equity category. The correct treatment depends on the database's stated purpose, not on a universal rule.

Row three: what counts as African

A company can be incorporated in the United States, founded by Africans, headquartered in Lagos, sell across several countries and employ a remote team. A tracker must choose a geographic test.

Possible rules use headquarters, primary operations, founder origin or a judgment about market focus. Acquisitions can move headquarters while the operating company remains. The choice affects companies with global structures more than local single-entity firms.

Country rankings need the same caution. Assigning a pan-African round to one headquarters concentrates the table. Splitting it by revenue or intended use would require data the announcement rarely provides.

Use the tracker's own country field and disclose its basis. Do not move deals between countries to fit a preferred narrative.

Row four: announcement date or closing date

A round may close in tranches and be announced months later. A tracker can record the announcement date because it is observable, or seek the legal close date. Neither choice maps perfectly to cash received.

An annual comparison needs the same rule in both periods. Live databases can revise earlier months after discovering a deal. Save the extraction date and expect totals to change.

Acquisitions and exits require another boundary. The Big Deal states that its cited H1 funding total excludes exits. A publication mixing acquisition consideration with primary capital would answer a different question.

Row five: disclosed value

“Raised $10 million” appears straightforward until the announcement combines equity, debt and an earlier unannounced tranche. A facility may be available over years. Local-currency amounts need an exchange rate and date for a dollar database.

Trackers may estimate undisclosed rounds from regulatory filings or reliable reporting. Others count the deal and leave value blank. Estimated and confirmed amounts should not look identical in a table.

Mega-deals amplify each decision. One classification can move a half-year total materially while leaving the underlying transaction unchanged.

Put totals in the final column

Build a crosswalk before comparing publishers. Rows should include minimum size, instrument treatment, geographic rule, event date, currency conversion, estimates, exits and source cut-off. The totals belong after those rows.

If definitions align closely, differences can point to missing deals or timing. If they do not, report both as separate views. Averaging them has no statistical meaning.

Company count also needs a label. TechCabal reported 146 disclosed deals, while The Big Deal counted 190 ventures above its threshold. A venture can complete more than one deal, and a deal can have several components. The nouns differ.

MarketBrief will cite a tracker beside every total and avoid constructing a trend from one publisher's current period and another publisher's base period.

What each view can answer

A broad deal database can show announcement flow and sectors attracting attention. A thresholded venture count can show how many companies cleared a meaningful capital level. An equity-only series can inform dilution and risk-capital analysis. A debt series can reveal asset finance and lender appetite.

No single total proves that fundraising became easier. Access depends on distribution, terms, stage and the companies that did not close. A complete market census is unavailable because private deals can remain private.

The two H1 2026 headlines are close enough to tempt a merged figure. Keeping them separate preserves more information: independent trackers observed a similar order of magnitude while describing the market with different units.

Sources

• TechCabal Insights, 3 July 2026: $1.44 Billion Raised in the First Half of 2026
• Africa: The Big Deal, 7 July 2026: June saves the day: H1 2026 funding review