A/B Testing Affiliate CTA Placements: What Actually Converts

Selin, a food creator based in Eskisehir, Turkey (52K Instagram, 18K TikTok), spent eight weeks testing the same affiliate coupon in three different placements. The result was sharper than she expected.

Test Setup and Methodology

Selin published 3 Reels per week. On Monday-Tuesday, the CTA appeared in the first line of the caption: "Restaurant owner? thMenu coupon: SELIN20 (link in bio)." Wednesday-Thursday it moved to the last line. Friday-Saturday it was removed from the caption entirely and lived only as a Linktree sticker in her bio. Sundays were reserved for analytics.

Each variant carried a distinct UTM tag: utm_content=cap_first, cap_last, bio_sticker. Linktree analytics and the thMenu affiliate dashboard were cross-checked weekly. Total: 8 weeks x 6 videos = 48 Reels, around 620K total views.

Results: First Line Won by 4x

Eight weeks of data produced a clean hierarchy. Click-to-signup-to-paid conversion rates:

- Caption first line: 2.8% final conversion (47 paid signups from approximately 206K view share)
- Caption last line: 1.1% final conversion (18 signups)
- Bio sticker: 0.7% final conversion (11 signups)

First line outperformed last line by 2.5x and bio sticker by 4x. Why? On Instagram, only around 22% of viewers tap "see more" to expand the caption — so anyone reading the last line is already a minority. Bio stickers require two extra steps: navigate to profile, then tap the sticker.

Practical Takeaways and Pitfalls

Your primary CTA belongs in the first line of the caption, but it shouldn't live there alone. Selin's best week combined first line + last line + bio sticker, reaching 3.4% conversion. Think of placement as a ladder, not a single rung.

Pitfalls: a hard CTA in the first line breaks your storytelling rhythm. Selin developed a hybrid sentence format — "Today's kebab story: SELIN20 -> link in bio" — where the CTA fuses with the hook. Also, Instagram may throttle reach for posts that pattern-match as promotional, so cap aggressive posts at 2-3 per week.

FAQ

Is 8 weeks enough for statistical significance? Yes — 48 videos and 620K views push p-value below 0.01.

Does this hold for TikTok? No. TikTok captions render shorter; pinned comments and bio outperform. Run a separate test.

Which tools did she use? Linktree Pro analytics ($5/mo), Instagram Insights, the thMenu affiliate dashboard, and a Google Sheets log.

Found this helpful? Share it.