How One Company Turns 200,000 Ad Comments into 109,000 Product Clicks

Most brands treat ad comments as a moderation problem. Something to monitor, occasionally respond to, and mostly ignore. One European DTC company treats them as a revenue channel — and the numbers back it up.

This company manages 31 Facebook and Instagram pages across 10+ consumer apps spanning wellness, parenting, photography, fashion, and education. Over the past 8 months, they've processed more than 200,000 incoming comments using AI-powered reply automation. The result: 109,000 clicks to their product pages — from comment replies alone.

We analyzed their full dataset to understand what's working, what isn't, and what other brands can learn from their approach.

The Numbers

Here's the all-time summary across their 31 pages:

The headline number: every shortlink placed in a comment reply generates 2.53 clicks on average. But that's a floor, not a ceiling, because links posted months ago are still collecting clicks today.

Growth Curve: From 10K to 38K Comments per Month

Comment volume grew nearly 4x in six months, from under 10K in July 2025 to nearly 38K in January 2026. This isn't organic growth. It's the result of scaling ad spend across more pages and more products.

What's notable is that AI reply volume stays relatively stable even as comment volume spikes. Not every comment gets a reply. Some get moderated, some don't warrant a response, and some fall below the confidence threshold. The system is selective by design.

The flywheel here is straightforward: more ads → more comments → more replies → more clicks → more revenue to fund more ads.

The Click Compounding Effect

This is the most interesting finding in the dataset.

Look at the divergence: shortlink creation stays relatively flat at 3,000–6,000 per month, but clicks grow from 12K to 33K, nearly tripling while link creation barely changes.

Why? Because shortlinks in comment replies persist. An ad posted in September still gets views in January. New people see the comment section, read the AI reply, and click the link. Old posts keep generating traffic without any additional effort.

This is the compounding effect: each month's shortlinks add to the cumulative base. By January 2026, this company had 43,000+ active shortlinks across thousands of posts, all silently collecting clicks.

Cohort analysis confirms this pattern. Links created in December 2025 have already accumulated 6.08 clicks each on average, while November 2025 links sit at 4.61. The longer a link exists, the more clicks it collects:

December links outperform January links because they've had more time to accumulate. This is a compounding asset: the value grows with time, not just with activity.

Not Every Page Is Equal: Clicks by Brand Category

Not all pages perform the same. When we break down clicks per shortlink by page type, a clear pattern emerges:

The top performers share a common trait: they're persona-driven pages, pages that operate as individuals rather than brands. A "Photography Persona" page (3.73 clicks/link) outperforms the brand-style "Photo Animation" page (2.74 clicks/link) for the same product.

The "Parenting Challenge" page, structured as an interactive challenge rather than a brand, leads the entire portfolio at 9.47 clicks per shortlink. That's nearly 4x the portfolio average.

The insight is clear: people trust people more than brands. When a reply comes from what appears to be a real person's page, readers are far more likely to click through. This isn't unique to comment replies. It's a well-documented pattern in social media marketing, but the data here quantifies the difference.

The 41/33/26 Split

Not every comment gets a reply. Here's what happens to every 100 comments that come in:

41 get an AI reply. These are substantive comments where the AI has high enough confidence to generate a relevant response, often including a shortlink to a product page.

33 get moderated. These are hidden: spam, scam mentions, emoji-only comments, competitor promotions, or content that violates the page's moderation rules. Hiding these comments isn't just about brand safety; it improves the quality of the visible comment section, making the remaining replies more prominent and trustworthy.

26 pass through without action. These might be simple reactions ("nice!"), comments in unsupported languages, or cases where the AI's confidence score fell below the 0.95 threshold.

The moderation and reply functions work together. By hiding a third of all comments (mostly spam and low-quality noise), the remaining comment section is cleaner, and the AI replies with product links stand out more. It's a two-part strategy: clean the room, then fill it with value.

What Makes Their Setup Work

This company runs 26 AI agents across their 31 pages. Here's what they have in common:

• Cost-optimized AI model — The AI model powering their replies strikes a balance between quality and cost, good enough to produce genuinely helpful responses, cheap enough to sustain 80K+ replies per month without breaking the bank.

• Smart link insertion — Links are appended "when relevant," not on every reply. The AI decides whether a product link fits the context. This avoids the spammy feel of link-stuffing every response.

• Aggressive moderation — One-third of all comments get hidden. Their rules target scam mentions, spam patterns, emoji-only comments, and competitor promotions. The result is a visibly cleaner comment section.

• Full automation — 22 of 26 agents run in fully automatic mode, where every generated reply gets posted immediately. The remaining 4 use hybrid mode with a 0.95 confidence threshold: high-confidence replies auto-publish, the rest get queued for human review. At their scale, mostly-automatic is the only way to keep up with 30K+ comments per month.

• Every-post coverage — All agents respond on every post type: ads, organic, everything. No filtering by post type. If someone comments, they get a reply.

The ROI Math

Let's do the math on what this costs versus what it delivers.

Total spend over 8 months:

• Subscription: $299/month x 8 months = $2,392
• (They started on a cheaper plan; actual spend is slightly less)

What they got:

• 109,251 clicks to product pages
• Cost per click: $0.022

For context, the average Meta (Facebook/Instagram) cost-per-click for ads ranges from $0.50 to $2.00+ depending on industry and targeting. This company is getting clicks at roughly 1/25th to 1/90th the cost of paid ads.

There's an important caveat: these are clicks to product landing pages, not guaranteed conversions. But the traffic is warm — these users already engaged with an ad (they commented on it), read a relevant AI reply, and chose to click through. That's a higher-intent visitor than a typical ad click.

Even at a conservative 1% conversion rate on those 109K clicks, that's 1,092 conversions from a $2,400 investment. At a modest $20 average order value, that's $21,840 in revenue. The actual numbers are likely higher, but without conversion tracking connected to the shortlinks, we can only estimate.

How to Replicate This

Whether or not you use the same tools, the principles from this data are transferable:

1. Treat comment sections as a traffic channel, not just a support queue. Every comment is an opportunity to surface a relevant link. Most brands leave this traffic on the table entirely.

2. Use trackable shortlinks in replies. Without tracking, you can't measure what comments actually drive. UTM parameters and shortlinks turn anecdotal "we think comments help" into measurable data.

3. Moderate aggressively. Hiding spam and low-quality comments isn't just about brand safety; it makes your good replies more visible and more clickable. The data shows a clear correlation between moderation rate and click-through rate.

4. Use persona pages for higher engagement. Persona-driven pages consistently outperform brand pages on clicks per shortlink. People trust replies from "people" more than replies from brands.

5. Set quality thresholds high. This company uses a 0.95 confidence threshold, meaning 1 in 20 potential replies gets held for review. Bad replies destroy trust faster than no reply at all. It's better to reply less and reply well.

6. Think in compounding terms. Every reply with a link is a permanent asset. Unlike ads that stop working when you stop paying, a comment reply sits there indefinitely, collecting clicks from future visitors. The ROI improves with time.

This client uses ReplyZen to manage their comment automation at scale, but the strategic principles apply regardless of the tool. The key insight isn't about AI or automation. It's about recognizing that comment sections are an underpriced traffic channel that most brands completely ignore.

Not every company takes the selective approach. In our second case study, a $100M/year gov-tech company went the opposite direction: one page, one agent, 96% reply rate, fully automatic. Their ROAS went from 1.4x to 1.7x. Different strategy, same underlying principle.

FAQ

How many comments do you need before this strategy works?

You don't need 200K comments to see results. The compounding effect starts with your first shortlink. Even at 1,000 comments per month, if 40% get replies with links and each link generates 2-3 clicks, that's 800-1,200 free clicks per month. The strategy scales linearly with volume.

Does this work on Instagram too, or just Facebook?

The AI reply and moderation strategy works on both platforms. However, the shortlink-in-reply approach is Facebook only. Instagram doesn't support clickable links in comments — any URL posted in an IG comment appears as plain text. This company runs AI agents on both Facebook and Instagram, but the 109K click numbers come exclusively from Facebook comment replies. On Instagram, the value is in engagement, moderation, and driving traffic through comment-to-DM automations or directing users to the link in bio.

What's the difference between a comment reply link and a comment-to-DM strategy?

Comment reply links are public, visible to everyone who views the post. One reply can generate dozens of clicks over time from different viewers. Comment-to-DM strategies (sending a private message when someone comments a keyword) are one-to-one. Each DM reaches only the original commenter. This company uses both: 374 DMs triggered from 40K keyword matches, alongside 43K public shortlinks generating 109K clicks. The public approach scales better.

How do you prevent Facebook from flagging automated replies with links?

Three factors help: First, replies are contextually relevant. The AI reads the comment and crafts a specific response, not a generic template. Second, the reply timing mimics human behavior, with responses staggered, not instant. Third, links are placed selectively ("when relevant"), not on every single reply. Facebook's spam detection targets repetitive, templated content. Context-aware, varied replies with natural timing don't trigger the same flags.

Data covers July 2025 through mid-February 2026. February figures are partial-month. All data is anonymized; no brand names, page URLs, or team member names are disclosed.