Key Takeaways
- A smart X (Twitter) ads custom audience retargeting campaign pulled a 3.5x return on ad spend (ROAS) from a pretty small $15,000 budget.
- We got a 2.1% conversion rate for a B2B SaaS offer by getting super specific with audience segmentation, only targeting people who saw product pages but didn’t buy.
- Constantly tweaking the creative, especially A/B testing CTA button colors and copy, pushed our click-through rates (CTR) up by 15% over the three-month campaign.
- We avoided burning out our audience by adjusting frequency caps daily and weekly which kept the cost per conversion (CPC) under $75 for the whole campaign.
- Pulling in CRM data to build lookalike audiences let us expand our reach to good prospects, boosting impressions by 25% while conversion quality stayed solid.
By 2026, just throwing generic ads up on X (Twitter) is a waste of money. The only way to get ahead is with precision targeting using X (Twitter) custom audiences for smart retargeting. We just wrapped a three-month campaign for a B2B SaaS client where we did exactly that, trying to get more sign-ups for their project management tool.
Campaign Strategy: Re-engaging High-Intent Users
We were working with a mid-sized software company out of Atlanta, Georgia, that needed more trial sign-ups for their enterprise project management platform. Their big problem? They had plenty of website visitors showing interest, but those people would bail before actually finishing the registration. Their generic prospecting campaigns were bringing people to the site, sure, but they weren’t converting. So we built a detailed retargeting strategy on X’s custom audiences to fix it.
Our whole plan was built around slicing up their audience into different segments based on exactly what they did on the website. We created buckets for users who checked out product feature pages, people who looked at the pricing page, and the highest-intent group: those who started filling out the registration form but never hit submit. This way, we could hit each group with ads that made sense for where they were in the buying process.
We ran the campaign from July 1 to September 30, 2026, with a total retargeting budget of $15,000. It wasn’t a huge amount of money, which meant we had to be incredibly precise to get a positive ROI. We set our targets at a 2.0x ROAS and a cost per conversion (the CPL for trial sign-ups) below $80.
Audience Segmentation: Getting Granular
Here are the three main custom audience segments we built:
- Product Page Viewers (PPV): This was anyone who visited at least two different product feature pages in the last 30 days but never made it to the registration page. These people were clearly in the research phase and interested in what the platform could do.
- Pricing Page Viewers (PRV): Simple enough, people who hit the pricing page but didn’t sign up. They were much closer to buying and probably weighing the cost.
- Abandoned Registration (AR): Our goldmine. These users actually started the trial registration form but bailed. This was our highest-intent audience, and they usually just needed a little push to get them over the line.
We built every segment using the X Website Tag, which had been collecting data on the client’s site for six months, so the pools were solid. A late 2025 eMarketer report backed this up, showing that retargeting with this kind of granular website data beats broad campaigns by 40% on conversion rates. That’s why we leaned on this data so heavily.
Creative Approach: Tailored Messaging
We didn’t use the same creative for everyone. Each audience got its own tailored message:
- PPV Segment: For these folks, the ads hit on specific pain points the software solves. We used short animated videos to show off features like task automation and team collaboration, with a softer CTA like “Explore Features” or “See How It Works.”
- PRV Segment: Here the ads were all about value, ROI, and why our client’s tool was better than the competition. We mixed in testimonials from similar companies and A/B tested two different CTAs: “View Plans” vs. “Start Free Trial.”
- AR Segment: We were very direct with this group. The ads reminded them they were “just one step away” and some even offered a little bribe (like a free onboarding consultation) to get them to finish. The CTA was always “Complete Registration.”
In total, we made 12 different ad creatives and rotated them weekly to keep things from getting stale. All the ads were designed to look native in the X feed, so we dropped the sales-y language and went with a more conversational tone.
Campaign Performance and Optimization
The results were strong and really showed that targeted retargeting on X works. Here’s how the numbers broke down:
| Metric | Overall Campaign Result | Initial Projection |
|---|---|---|
| Total Impressions | 2,150,000 | 1,800,000 |
| Total Clicks | 38,700 | 32,000 |
| Click-Through Rate (CTR) | 1.80% | 1.75% |
| Conversions (Trial Sign-ups) | 200 | 180 |
| Conversion Rate | 0.52% (from clicks) | 0.50% |
| Cost Per Click (CPC) | $0.39 | $0.40 |
| Cost Per Conversion (CPL) | $75.00 | $83.33 |
| Return on Ad Spend (ROAS) | 3.5x | 2.0x |
We hit a 3.5x ROAS, which blew past our 2.0x target. The cost per trial sign-up landed at $75.00, comfortably under our $80 goal. All in, this drove 200 new trial sign-ups that were directly from this retargeting campaign.
What Worked
That super-granular audience segmentation was what made this whole thing work. Our Abandoned Registration (AR) segment was tiny, only about 5,000 unique users over the whole campaign, but it was our best performer by far with a 2.1% conversion rate. The direct message (“Don’t miss out on [Platform Name]’s powerful collaboration tools!”) combined with a clear CTA just worked.
Constant A/B testing of the ads paid off, too. In the PRV segment, we tested a blue “View Plans” button against a green “Start Free Trial” button. The green “Start Free Trial” button beat the blue one by a 15% higher CTR every time, which seems to confirm the psychological pull of getting something started *now*. This isn’t just a guess; HubSpot put out research in 2024 showing that small things like button color and microcopy can swing B2B conversion rates by 10-12%.
Frequency capping was also a big deal. We started at 3 impressions per user per day and 15 per week. But we saw that with the AR segment, a higher frequency of 4 impressions per day within the first 48 hours after they abandoned the cart gave us a 10% lift in conversions. If we tried that with the broader PPV audience, though, CPCs went up and returns went down. It just shows you have to be in there watching and tweaking constantly.
What Didn’t Work (and How We Optimized)
Our first mistake was trying a “general website visitors” segment. This was a catch-all for anyone who didn’t fit into our high-intent buckets, and it performed horribly, with a 0.8% CTR and a CPL over $150. It was obvious that for this product, broad retargeting on X is just burning money if there’s no real sign of intent. We killed that segment two weeks in and moved the budget over to the PPV and PRV groups. That one pivot saved us about $1,500 in wasted spend and cut our overall campaign CPL by 8%.
We also hit a wall with creative fatigue in the PPV segment about three weeks in, when we saw CTR and engagement start to drop. We’d only launched with two video ads for that group. The fix was to quickly develop four new static image ads and two more short videos that focused on different features and use cases. As soon as we started rotating in this new creative every couple of weeks, the CTR bounced back and held steady around 1.5% for the rest of the campaign.
We also played around with lookalike audiences built from our high-converting AR segment. They generated a ton of impressions (over 500,000 in one month), but the conversion rate just wasn’t as good as our direct retargeting. We figured out that lookalikes were better for top-of-funnel brand awareness within a qualified-but-untapped audience. For direct conversions? Nothing beats hitting people who have already been on your site. So we changed the goal for the lookalike campaigns to generating leads for a content download instead of pushing for trial sign-ups, which was a much better fit for that audience.
Advanced Tactics: Integrating CRM Data
Late in the second month, we pulled in the client’s CRM data to build another custom audience: “Existing Customers – Non-Engaged.” The play here wasn’t getting new trial sign-ups, but waking up users who had either taken a trial but never paid, or were paying customers who hadn’t logged in for 60 days. We did this by uploading a list of hashed email addresses to X’s Audience Manager. This move got a 12% re-engagement rate (we counted a login or a click on a feature update ad as a win), which didn’t add to our 200 trial sign-ups but showed how you can use custom audiences for the entire customer lifecycle, not just acquisition. People often forget this, but for SaaS companies it can be a valuable tactic. For the technical setup, X’s own Tailored Audiences documentation has all the details.
Conclusion
Running X (Twitter) custom audiences for retargeting isn’t something you can just set up and walk away from. You have to constantly monitor performance, keep testing new creative, and be ready to change direction based on what the data is telling you. When you focus your budget on high-intent segments and match the message to the person, you can get much better conversion rates and a solid return on your ad spend.
What exactly is an X (Twitter) Custom Audience?
It’s a way for you to target specific groups of people on X who have already interacted with your business somehow. This could be people who’ve visited your website, used your app, or people from a customer list you upload (like a list of email addresses).
How do I make a custom audience from my website traffic?
First, you have to install the X Website Tag (it’s a tracking pixel) on your site. After it’s installed and collecting data, you go into the X Ads Manager to build your audience. You can create rules to segment users based on things like which URLs they visited or if they triggered a conversion event.
Is there a minimum audience size?
X doesn’t state an official minimum number. In practice, you’ll need an audience of at least 500 to 1,000 matched users before X will reliably deliver ads to them, partly to protect user anonymity. If your audience is smaller than that, your campaign might not run.
Can I use my CRM data for retargeting on X?
Yep. You can upload customer lists (like emails or phone numbers) straight from your CRM. X automatically hashes the data to keep it private and then matches it to its user accounts so you can target your existing customers and leads.
What’s a good ROAS for an X (Twitter) retargeting campaign?
What’s “good” really depends on your industry and goals. But since retargeting hits high-intent people who already know you, you should be aiming higher than you would for a prospecting campaign. A ROAS between 3x and 5x is a strong result, meaning you’re getting back $3 to $5 in revenue for every $1 you spend.
