What mapping 1,145 African startups taught us about distribution

This lands. We kept optimizing the "matching" side of an early tool and ignored that founders mostly needed to be findable. Same trap shows up outside directories: you build the thing that solves your thesis, then the side channel (a listing page, a public profile, a simple shareable artifact) quietly does more distribution than the feature you obsessed over.

Useful reframe for anyone shipping a two-sided discovery product: pick which side is starving for attention first, then make that surface permanent and indexable before you add more matching logic.

The distinction between ‘finding something’ and ‘being findable’ is clean. A lot of early tools solve the first and completely miss the second. Treating the listing page as a first-class surface instead of a side effect feels like the right correction. Curious whether you’ve seen founders start treating their profile differently once they realise it can rank and bring inbound on its own.

The distinction between “finding funding” and “being discoverable” is really interesting. I think early-stage founders often focus so heavily on acquisition that they overlook the value of creating assets that can keep bringing discovery over time.

The individual startup pages seem especially powerful because they turn each listing into a potential entry point from search, rather than requiring someone to start from the directory homepage.

I’d be curious whether you’ve seen a difference in conversion or engagement between founders who actively optimize their listing and those who simply create one. That could reveal whether the listing itself is the distribution channel or whether the real advantage comes from how it’s used.

The listing page finding is the right one, and there is a repeatable weekly supply of the exact founders who need it.

Mathesis Analytics, Lagos, raised from First Ally Capital in July and was covered by seven outlets. Its Dealroom entry has an empty website field and mathesisanalytics.com does not resolve. Credit infrastructure touching millions of borrowers, and no page the company controls.

KilimoChills, solar cold storage in Kenya, made two finalist lists six days apart this month, ASME ISHOW and the ACEP energy challenge. Search the name and you get only those two lists. The founder, Tracy Achieng, is named on the second one, so anyone who reads the first cannot reach her at all.

That is the seam worth mining. Engineering and climate shortlists publish every few weeks, full of ventures with real validation and no site, because those founders build hardware rather than landing pages.

I hand-build prospect lists for founders. Want the rest? Free, no strings.

The listing becomes much more useful if founders can see the exact searches and referral pages that brought people there. “You were discovered” is nice; “three investors found you through fintech startups in Kenya” tells them what to improve next.

1,145 profiles is a useful proof of discovery, but I’d separate indexed pages from maintained pages in the next update. A simple cohort view — how many profiles get an edit or inbound click within 30/90 days — would tell you whether the listing is a durable surface or a one-time SEO artifact. The funding matcher can then be judged against the same signal: does a match lead to a profile visit or a founder update? That connects the two sides without forcing one metric to stand in for the whole product.

What decides this is decay, not discovery. A directory loses its value the moment a third of the listings are stale, so the real product is the update loop: a quarterly nudge that makes a founder reconfirm stage, headcount, and revenue band, with a visible last verified date on every page. Worth testing which side pays too, because founders will pay a little for visibility, but investors and corporate partners pay real money for a filtered, current, verified list, and I would rather sell that than a listing fee.

Same lesson from the other side: we described the product perfectly on the landing page and got nothing, because the people with the problem weren't searching — they were already talking about it somewhere else. The starving side for us was presence, not pages. A listing ranks over months; a useful reply in a live thread reaches the right person in minutes. Only one of those starts the loop early.

This is a measurement boundary problem you solved by watching behavior instead of assumptions. You measured "matching success" (the feature) and missed "being found" (the outcome). Founders didn't abandon the matching engine - they just discovered that being indexed separately solved a different, more urgent constraint. The insight is that matching and discoverability are two completely different measurement systems competing for the same real estate. A founder's distribution problem isn't "how do I find what's relevant" but "can the person looking for me actually find me." Once you switched to measuring who got found (through search, through linking, through being crawled), everything else rearranged around that metric. The matching engine solves a search problem. The listing page solves a findability problem. Most founders face the findability one first.

A listing becomes much more valuable when it is treated as a real entity page rather than a generic directory entry. I’d add structured facts, regular updates and a visible “last verified” date so search engines and visitors can distinguish active startups from abandoned profiles. It would also be interesting to compare how many visits arrive from category searches versus searches for the startup’s exact name.

Your finding matches something I ran into from the opposite direction, and it might be worth more than it looks.

We built a sample of 139 US small business websites by taking the top organic results for local service queries in twelve cities. The single most laborious step was throwing things out: directories, marketplaces, review sites and news outlets, because that is what the top of local search is mostly made of. We wanted the businesses' own sites and had to dig for them. I didn't record the ratio, so I won't quote one — but the filtering was the work, not a footnote to it.

Read from the business owner's side, that is your thesis restated: the page where people find you is very often not your site. The listing does the distribution whether or not anyone planned it that way.

One refinement on "a real backlink". I'd separate the two things it does, because they are worth different amounts. As a link passing authority, a directory listing is close to worthless and everyone in SEO now says so. As a second page that exists, gets indexed and ranks for the company's own name, it is genuinely valuable to an early-stage founder who has nothing else ranking. Selling it as the second thing is honest and holds up; selling it as the first invites an argument you'd lose.

And a practical one, since your listings compete with the startups' own sites for the same query: keep the listing pages fast. In our sample the median business site took 8.9 seconds to render its main content and 3% met Google's 2.5-second threshold. A listing page that loads instantly doesn't just rank — it converts the visitor who would otherwise bounce off the founder's own homepage.

The pattern you have hit is the one that makes directories defensible, and it is worth naming because it changes what you build next. The matching engine is the product. The listing pages are the distribution. Those are two different businesses and only the second one compounds.

Two things I would watch, one risk and one opportunity.

The risk: 1,145 pages is roughly the size where scaled content demotion stops being theoretical. What protects a directory is that every page is a real entity with real data, which yours are. What breaks it is the tail of stubs, startups that got listed and never filled anything in. Those pages do not just fail to rank, they drag the domain, because Google assesses the pattern across a template rather than page by page. The fix is unpopular and it works: noindex the thin ones and keep the substantial ones indexed. 400 good pages will outrank 1,145 where 700 are empty.

The measurable version of that: pull Search Console, filter to the listing URL pattern, and look at how concentrated impressions are across those pages. My guess is a long tail sitting at literally zero. That concentration ratio is the health metric for a directory, not the total listing count, and it is the number I would put on the wall instead of 1,145.

The opportunity is the thing you have already stumbled into. Because founders want to be found, they will maintain their own page, which is free content maintenance at a scale you could never staff. That is the actual moat. I would design for it deliberately rather than let it happen by accident: show each founder how many people viewed their listing this month and what their profile is missing, and a static row becomes something they come back and update. Freshness on a page somebody personally cares about is very hard for a competitor to copy.

Which is what I would want to know. Of the 1,145, how many have been edited by the founder at any point after creation? That ratio tells you whether you have a flywheel or a large static database that happens to rank.

Interesting that the listing page became the stronger distribution surface.

Curious whether founders started valuing the directory differently once they realized their profile itself could bring discovery.