The Monday after South Africa’s startup ecosystem rose 31.3% in StartupBlink’s 2026 index, a product team should replace the celebratory ranking slide with a cohort chart showing whether last quarter’s customers returned. The ranking describes the environment around the company; retention reveals whether the company has built something customers still need.
In 2004, Eric Ries and the IMVU team in Silicon Valley faced a harsher version of that distinction. They had built and improved a product, tracked activity, and expected their work to produce growth. The numbers they watched suggested movement. The business still lacked the customer response they needed.
Ries later documented the episode in The Lean Startup. The uncomfortable finding came from examining customer behaviour in cohorts rather than taking comfort from totals. Product changes were shipping, yet the measurements that mattered were barely moving. More work had created more activity without proving more value.
The ranking changes the story around the company
A stronger ecosystem ranking can matter. It may affect how investors scan markets, how international partners interpret South African startups, and how founders speak about the conditions around them. A 31.3% rise deserves attention.
It cannot answer the questions waiting inside Monday’s product meeting.
Did customers who first paid last quarter return? Did teams use the product after onboarding? Did the AI feature solve the job that brought them in, or did it produce an impressive first session followed by silence? Did revenue grow because customers stayed, or because the team kept replacing those who left?
Those questions sit closer to payroll and runway.
I have seen how easy it is to confuse a favourable external signal with evidence that the product is working. A founder in Johannesburg may now have a stronger story for an investor in Berlin or a prospective partner in Accra. That story can open a conversation. It does not keep a customer.
The distinction also matters when a ranking arrives during a difficult quarter. A founder can feel pressure to perform the optimism around the ecosystem while privately deciding whether to hire another engineer, reduce scope, or give one customer segment another month. The ranking and the company can both be moving upward at different speeds. Sometimes one rises while the other remains flat.
That tension sits behind what happens when South Africa’s startup ranking rises but your runway does not.
A cohort chart removes the flattering explanation
A total customer count lets several different realities occupy one number. Fifty customers could mean forty-five stayed. It could also mean most left and were replaced by new sales.
A cohort chart separates those realities.
Group customers by the period when they started, then track the behaviour tied to the product’s value. For a workflow tool, that may be a completed process. For an AI product, it may be a repeated task completed with an accepted output. For a SaaS product sold to small teams, it may be continued use by more than the person who approved the purchase.
The useful event should represent value received, not a click that happens because an email reminder arrived.
Then look at the last few cohorts without blending them into one average. If each new group returns less often, acquisition can hide the decline for a while. If a smaller cohort retains better after a specific product change, that is evidence worth investigating. It still needs explanation, especially when the sample is small, but it gives the team somewhere concrete to look.
This is the same discipline behind validating demand before committing scarce build time. Ruth’s refusal in this account of AI product validation mattered because a customer decision revealed more than internal enthusiasm could.
Retention turns celebration into a decision
The point of replacing the slide is not to dampen the team’s mood. It is to convert a broad signal into a company-level decision.
Suppose the previous quarter’s customers returned and repeated the core action. The team can ask what those customers share, which promise brought them in, and where expansion is justified. The ecosystem news may then strengthen a story already supported by product behaviour.
Suppose they did not return. The next move is smaller and less glamorous: speak to the lost customers, inspect where usage stopped, and test one explanation before adding another feature. A team with limited runway cannot afford to answer weak retention with a larger roadmap.
Ries’s IMVU story remains useful because the team’s effort was real. The product work was real. The activity was real. Cohort analysis exposed the missing connection between shipping changes and changing customer behaviour.
That is the bridge to a South African founder on Monday morning. The ecosystem ranking can change how the market sees your location. Only customer behaviour can show whether your last product decision worked.
Keep the ranking slide for the board appendix. Put the cohort chart on the meeting-room screen, select last quarter’s customers, and begin with the people who did not come back.
Comments
No comments yet.