Which channels should a deals platform keep funding?
Author of the evaluation and its analysis
Early 2026, on data from 2022 to February 2026
Web analytics, search console data, SQL, spreadsheet
The question
A deals platform for online sellers had grown for three years across about a dozen channels, YouTube first among them. Which channels had earned continued budget, and which had not?
The method
I followed each visitor through four steps: visit, signup, deal redeemed and, from mid-2025, a purchase on the partner's side.
Judging channels at the last step mattered. Signup rates were similar across the main channels, but redemption quality was not.
The purchase step was estimated, not measured. I applied an assumed low-single-digit partner purchase rate to redemptions, with a conservative and an optimistic case.
I compared spend per redemption, never per visitor, so cheap but empty traffic could not look efficient.
The evidence
| Channel | Reach trend | Quality after signup | Cost trend | What the evidence said |
|---|---|---|---|---|
| YouTube | Down roughly 30% in 2025 after driving most early traffic | Improving each year | Cost per redemption up roughly six to seven times from 2023 to 2025 | Built the product, now shrinking and costly |
| Search and AI-assistant referrals | Growing every year | Highest of the main channels | Spend about a tenth of YouTube's | Compounding without ongoing spend |
| Short-form video | Filled YouTube's gap in 2025, then collapsed in 2026 | About half YouTube's | Pay-to-play | Creator pool exhausted |
| Paid ads | Too small to judge | Strong on a tiny sample | Acquisition cost did not add up | No conclusion possible |
| Direct | Growing | Steady | No independent spend | Follows the other channels |
The result
Three findings stood out. Traffic doubled to 2024 and then flattened, so every later gain came from funnel quality. The signup rate nearly tripled between 2023 and 2025 with no extra visitors.
Search was the only channel growing in both volume and quality, though about three pages produced roughly three-quarters of its signups. AI-generated answers in search results were an emerging threat, because people could see a deal code without visiting the page.
And volume was too low to win exclusive deals from partners, so the platform's edge was curation and presentation, not inventory.