When qualified sign-ups disappear, repair the referral channel only after you understand what it was hiding. A partner can restore this week’s pipeline, but only customer discovery can tell you how to build the next one.
In January 2018, Jonah Peretti’s BuzzFeed faced that problem at a much larger scale. Facebook announced that its News Feed would give more weight to posts from friends and family. Publishers that had built distribution around the platform could no longer assume the same flow of readers would continue.
The outcome was still uncertain. BuzzFeed had become exceptionally good at producing stories people shared on Facebook, and Facebook had benefited from that material. The relationship had worked until one side changed the rules.
The channel worked until it became the strategy
Facebook’s announcement, published by Adam Mosseri in the company’s Newsroom, explained that public content from businesses, brands and media would receive less emphasis. The change did not single out BuzzFeed, but it altered the ground beneath publishers whose audience acquisition depended on the Feed.
BuzzFeed could improve its posts, speak with Facebook and adjust its publishing tactics. None of those actions could restore control over distribution. Facebook owned the ranking system, the audience relationship and the decision about what appeared next.
That distinction matters when a founder opens the dashboard before lunch and sees qualified sign-ups below their usual level.
Suppose a software company in Accra receives most of its trials through a partnership with an agency in Berlin. The referrals convert because the agency has already established trust. Then the introductions slow. The immediate instinct is reasonable: call the partner, inspect the handoff and find out what changed.
The dangerous assumption comes next: if the relationship is repaired, the underlying problem is solved.
It may only mean the company has recovered access to customers it still does not know how to find.
Separate a broken handoff from borrowed demand
I would start with the individual sign-ups, not the weekly total.
Which customer profiles disappeared? Which use case did they bring? What did the partner say before sending them? Did those prospects arrive ready to buy, or did the product still have to persuade them?
A referral source often performs several jobs at once. It identifies the buyer, explains the problem, filters out poor fits and lends its reputation to the introduction. When the feed goes quiet, all four disappear together. The founder sees a traffic problem, although the missing asset may be trust, targeting or language.
This is why replacing one partner with three more can reproduce the same weakness. The chart looks diversified while every lead still depends on somebody else understanding the customer first.
I would call recent customers and ask them to reconstruct the moment before the referral. What were they trying to fix? What had they already attempted? Which phrase from the partner made the product worth considering? Where would they have looked if that introduction had never happened?
Those conversations are not a survey about whether customers like the product. They are a search for the path that created demand.
The same discipline appears in how a paid pilot changed Kweku’s product roadmap. A customer’s commitment can reveal more than a founder’s confident account of the market.
Run the repair and the investigation together
The choice between saving the partnership and learning independent distribution is false when runway is limited. Do both, with different time horizons.
Repair the handoff first. Confirm whether the referral link works, whether qualification criteria changed and whether the partner has shifted attention to another offer. Ask for a short conversation with the person who owns the relationship. Avoid promising discounts, custom features or exclusivity before knowing what failed.
At the same time, preserve the evidence. Export the referral cohort. Compare it with customers who arrived directly, through search, from an event or by word of mouth. Look for differences in company size, urgency, objections and time to activation.
Then test one route the partner does not control. Use the language customers gave you in a focused article, a direct outreach message or a landing page for one use case. Send it to a small, relevant audience. The first goal is learning whether the message earns a reply from the same kind of buyer.
This is closer to testing a strategy before spending runway than launching a broad acquisition campaign. A small test can show whether the partner supplied reach, trust or the entire explanation of the problem.
Own the explanation before chasing the volume
BuzzFeed did diversify beyond Facebook, but the company’s later restructuring showed how costly platform dependence had become. The New York Times documented its January 2019 plan to cut about 15 percent of its workforce. One algorithm update did not explain every part of BuzzFeed’s business, but the episode made the distribution risk difficult to ignore.
The founder watching a quiet referral feed faces the same mechanism at a smaller scale. Borrowed distribution can grow a company quickly. It can also delay the work of learning who buys, what triggers the search and which words make the product credible.
By the end of Tuesday, I would want two things: a clear owner and next step for repairing the partnership, plus five customer conversations booked from the affected cohort. Restore the feed if possible. Use the silence to learn what the feed had been doing for you.
Comments
No comments yet.