A former Wieden+Kennedy strategist named the thing most brand documents cannot do: govern a decision. The same fortnight, African capital held flat, deal count fell 42%, and 84% of the money landed on 30 companies. The visible layer stopped paying.
A weekly intelligence brief from Base X Studio. Pan-African tech, global brand strategy.
African startups raised roughly $1.4bn in H1 2026, almost exactly matching H1 2025, while deal count fell 42% year on year. The same money is doing a different job. Rounds under $500k have fallen from 52% of all deals in 2021 to 19%, the $100k to $1m band dropped from 179 deals to 100, and grant funding collapsed from 27 awards worth about $20m in Q1 2025 to 15 worth about $4m. The tier where a business could be funded on a story is closing, and the non-dilutive path that used to bridge it is closing faster.
If your raise plan was traction plus story, the story half stopped paying. This quarter, stop refining the narrative and build the two or three evidence artefacts a concentrated investor actually reads: cohort retention, unit economics at current scale, a named repeat buyer. If your plan assumed a grant would cover the gap, replan now, because grant value fell roughly 80% year on year in Q1.
Average basket, around $32, across tens of thousands of delivered orders. Individual orders were profitable. Founded by Paystack alumni, funded on that credibility, partnered with Chowdeck from December 2025 to extend reach.
An owned chain, direct farmer buying, warehousing, fleet and proprietary software, carries fixed costs only frequency covers. GoLemon never reached the orders per zone per day the structure needed. It shut in July 2026 after failing to raise; support closed 2 August.
If you carry fixed costs across an owned chain, your real metric is orders per zone per day and not gross margin per order, because you can win the second and lose the company. This quarter, calculate the density your fixed costs actually require and compare it honestly to your current trend. If closing the gap needs a step change in frequency rather than a growth rate, narrow the geography until the density works before you spend anything more on reach. And whatever your pedigree is worth, do not price it as runway.
LemFi, with 2 million customers sending into more than 30 markets, will settle cross-border remittances over BVNK's regulated stablecoin rails, converting to local currency on arrival with no crypto step the sender ever sees. That invisibility is the story. Adoption is not arriving as a new consumer behaviour, it is arriving as an infrastructure swap behind a familiar product, market by market as regulation allows.
If you compete on remittance price, the floor just moved and it is not finished moving. This quarter, decide whether you are a price competitor or a trust competitor, because once settlement costs converge you cannot be both. If trust, fund the things settlement cannot supply now: the payout network's last mile, the recipient's experience, your compliance posture in a specific corridor. And copy the design decision, because LemFi's customers never learn the word stablecoin. Never make a user learn your infrastructure.
Zedcrest Group, a Nigerian investment banking, asset management and fintech conglomerate, acquired UK-founded Leatherback on undisclosed terms. Leatherback will run as an independent subsidiary keeping its brand and its management. Zedcrest first backed its $10m pre-seed in 2021. The decision worth reading is the non-absorption: the acquirer priced the brand and the team as part of the asset rather than as overhead to rationalise. Note the buyer too. Not a bigger fintech, a diversified financial group.
If a strategic exit sits anywhere in your five-year plan, this deal tells you what gets kept. Be honest about which side you are on. If your customer relationship lives in your brand and your people, that is priced, and treating brand spend as discretionary is a valuation decision you are making by accident. If your value is a licence, a rail or a merchant book any acquirer could operate, expect absorption. Either way, widen your list of plausible buyers beyond your own category.
A ranking that one company's removal inverts is a sentence about that company, not about a market. Six weeks ago this brief used a similar table to report the centre of gravity had moved to Nairobi. This one says Nigeria reclaimed the equity crown and Kenya fell. Both are honest readings of a single half-year. Neither is a thesis.
Before any geographic bet this quarter, an office, a hire, a raise location, run the one-line test: remove the largest single deal and see whether the ranking survives. If it does not, you are reading a company and calling it a market. Build the thesis on where your specific buyers are and treat the league table as trivia.
Martin Weigel argues organisations systematically underbuild the narrative infrastructure that makes business decisions coherent, favouring visible outputs like campaigns and platforms over the structural work, and he draws the parallel to physical infrastructure literally: US rail spending of $39 per head against Luxembourg's $625, the NHS's £15.9bn maintenance backlog. His distinction is the sharp part. A story is bounded, temporal, specific, an event that ends. A narrative does not stop, because it is the frame that decides what counts, what things mean, and how decisions connect. Brand onions, pyramids, wheels and mission statements, he writes, "describe badly; they do not govern", offering "aspiration dressed as structure" where testable propositional claims belong.
Open whatever passes for your brand document and look for one decision it would have stopped you making. If you cannot find one, you own a description rather than an infrastructure, and it is not doing the job you paid for. This quarter, convert the two or three things you actually believe into propositions specific enough to be wrong: who you will not serve, what you will not build, the price you will not go below and why. Put them where operating decisions get made, not in a deck.
Kasey Jones argues consultants fail to command premium pricing not because buyers are cheap but because offers are interchangeable, and when everything looks the same price is the only thing left to compare. Her fix loads the value into the opening. Diagnose the prospect's situation more precisely than they can articulate it themselves, in their own language. State outcomes as a concrete before and after rather than a deliverables list. Replace the credentials block with a specific point of view. Her cited results are self-reported, one client moving from $10k to $70k monthly in six months. The structural argument stands without them.
Open your last proposal and read the first page. If it describes you, your team and your track record, you handed the buyer nothing to compare except price. This quarter, rewrite the opening as a diagnosis: state the buyer's situation in their own words, more precisely than they would state it themselves, before you say a single thing about what you do or have done. A buyer who reads their own problem described better than they could describe it has already decided you are different.
Bina argues that pervasive deception across digital life, AI content outpacing detection, opaque supply chains presented as choice, near-constant scam risk in high-stakes transactions, has collapsed trust in mediated experience and pushed people back toward the physical. She splits markets in two: those that mediate between a person and their own body, the dashboards, scores and feeds, and those that enable direct experience. The inversion matters. The interface that used to be the product has become the thing buyers distrust, because the technology producing the dashboard also produces the fake.
In a market where anyone can generate a convincing case study, your dashboard, your report and your polished deck have stopped carrying proof, because the buyer knows a model could have made them. This quarter, move at least one part of your sales process into something unfakeable: put the buyer in a room with the person who would do the work, in front of a customer who will speak without a script, or on site where the thing actually operates. If your entire funnel is mediated, you are competing on production quality in a category where production quality just went to zero.
Tesla reports avoided emissions, 37 million metric tons of CO2 in 2025, a measure it built and then led on. Legacy carmakers cannot win a number a competitor authored.
Its predictive airbag fires 70 milliseconds before impact rather than 50 after, an estimated 2,700 fewer fatal crashes a year. A claim with a mechanism behind it.
The same car carries eight exterior cameras, interior cameras, microphones, GPS and cellular, and claims privacy. Claiming a word and earning it are separate acts.
Where a leader's claimed word contradicts its operating reality, that is not a complaint to make. It is a position available to whoever will back the word properly.
Work out which number your market judges you on, then work out who invented it. If a competitor authored the scoreboard, you play an away fixture every quarter and no amount of performance fixes a rigged comparison. This quarter do one of two things: define and publish a measure that reflects what you are genuinely best at and report it consistently, or name a word your category leader claims and cannot back, and take it. Both are cheap, and neither requires anyone's permission.
Martin Weigel, who ran strategy at Wieden+Kennedy Amsterdam, argues that organisations systematically underbuild narrative infrastructure, and that most of what the industry sells as strategy, the onions, pyramids, wheels and mission statements, "describe badly; they do not govern". A story is bounded and ends. A narrative does not stop, because it is the frame deciding what counts and how decisions connect. On its own that is a strong essay. Read against the funding data in this issue it becomes a market condition. African capital held flat while deal count fell 42% and 84% of it landed on 30 companies, which means the money stopped buying the visible layer and started buying evidence of the structural one. And GoLemon, founded by Paystack alumni with exactly the credibility that used to be enough, shut down with profitable orders and insufficient density. So open whatever passes for your brand document and look for one decision it would have stopped you making. If there is not one, you own a description, and the market has stopped paying for descriptions. This quarter, turn the two or three things you actually believe into propositions specific enough to be wrong: who you will not serve, what you will not build, the price you will not go below and the reason. Put them where operating decisions are made, then check next quarter whether a decision ever broke against them. If nothing ever does, the claims were never real.
The mechanic is portable and it takes an afternoon. A company claims a word, and the claim gets tested against its operating reality. Category Pirates ran it on Tesla claiming privacy while operating eight exterior cameras. Run it in your own market, the fintech that claims trust, the telco that claims local, the bank that claims access, in two steps: name the word your category leader owns, then find the decision that contradicts it. The gap between a claimed word and an operating reality is where your position lives, and nobody is guarding it.