Subscription Box Marketing: How Audience and Landing Page Segmentation Drove $29M for a Luxury Women’s Subscription Box

Case Study Summary

The Challenge

Advertising For a Unique Business Model

Scaling Within New and Existing Ad Channels

Increasing Ad Spend by 2x with 250%+ ROAS

The Solution

Insights from Success in Current Marketing

Omni-Channel Segmentation Strategy

Landing Page Segmentation Strategy

Customized Ad Copy and Creative Assets

Revenue-Centric Tracking Across Channels

Omni-Channel Budget and Bidding Optimization

Facebook, Instagram, and Pinterest Ad Segmentation

Google and Bing Search Ad Segmentation

Youtube and Display Ad Segmentation

The Results

102k+ New Member Subscriptions

3.4x Increase in Monthly Ad Spend

$29.4M In New Client Revenue with 268% ROAS

$6.9M Operating Profit from Ads with a 23.7% Margin

The Challenge

Our client, a Luxury Women’s Lifestyle Subscription Box service, had just raised a new round of funding and was looking to scale up their paid marketing efforts.

Historically, the client had primarily relied on organic SEO through their internal content team, social advertising, word-of-mouth, and referral programs to reach 325,000 members.

Advertising For a Unique Business Model

The client’s membership e-commerce platform offers women curated luxury subscription boxes (annually and quarterly) with up to 75% member savings across 100’s of luxury brands in beauty, skincare, jewelry, fashion, wellness, home goods, and other segments.

The interesting challenge and opportunity with this client came in that their business operated at the intersection of e-commerce, membership lead generation, and derivative-SaaS — functioning as a Product-as-a-Service (PaaS) company with similar LTV’s and business fundamentals to traditional SaaS companies.

Scaling Within New and Existing Ad Channels

While internal teams had found success running ~$180,000 in monthly advertising across Facebook, Instagram, Pinterest, and branded Google search campaigns, they struggled to find ways of profitably scaling their ad spend both within existing channels and across new channels.

The key goal was to execute an omni-channel advertising strategy that could profitably convert membership subscriptions as a sustainable way of growing the client’s business.

Increasing Ad Spend by 2x with 250%+ ROAS

We were tasked with the key challenge of doubling the client’s ad spend within 12 months (to $360,000 per month) while retaining a 250%+ ROAS for every dollar spent.

Given their aggressive targets, we had to identify opportunities both within and outside of existing advertising channels in order to succeed.

The Solution

Ultimately, we knew the keys to success would come in taking a segmented approach to their advertising across channels, ensuring that every touch point resonated with the right target audience, on the right channel, with the right content, and at the right time.

This meant designing a segmented omni-channel advertising strategy in which each channel’s efforts complemented the others in creating new awareness and converting memberships from the top to the bottom of the funnel.

Insights from Success in Current Marketing

Keyword Insights From Non-Branded SEO: With 63% of their organic traffic coming from non-branded search terms (i.e. people not specifically looking them up by name), we wanted to identify which search terms and pieces of content generated the highest conversion rates towards memberships. Ultimately, we found that the non-branded search terms with the highest membership conversion rates fell into 3 general categories:

1. Category subscription box searches (i.e. “fashion subscription box”) 2. Product-category-specific searches (i.e. “thickening mascara”)

3. Brand-specific searches (for specific brand items)

While category subscription box searches were a no-brainer for search advertising, the second segment uncovered a huge opportunity in launching segmented Google search campaigns that would show relevant ads for each product-category segment with clear value propositions for becoming a member. For the third segment, unfortunately we couldn’t bid or advertise on brand partner product terms due to the client’s partnership agreements. The ultimate goal of our search ads hypothesis was to get in front of people actively searching for product-categories that the subscription box offered — making it clear to the person searching that they could buy these products with up to 75% in savings by signing up for a membership subscription. Audience Insights from Branded Google Search Ads: While the client had only run a branded search campaign on Google, this campaign had been running for years spending anywhere between $10k up to $15k per month. These branded searches primarily came from people looking them up who found them through social channels, word-of-mouth, and referral programs. Therefore, by analyzing the demographics and audiences of these searches, we gathered key insights on the audience personas to target in Google search ads:

1. Women between 18 to 54 years old 2. Top 10 to 30% household income categories (and unknown income)

3. 48 affinity, in-market, and detailed demographic audience segments

Insights from Gaps and Opportunities in Social Media Advertising: In analyzing their social advertising accounts, we found that the audience personas from their branded search campaign aligned with the audience personas that converted on the client’s social media advertising. However, we found that the client was simply running all-in-one campaigns for each platform with very little audience or ad segmentation. This meant that they were using the same creative and ad copy for all their audience targets. Further, they were pointing every ad to their home page without any kind of segmentation of landing pages to match unique audiences and ad segments. They had also never leveraged the large volumes of varied creative used in their organic social marketing to be segmented in matching appropriate audiences in their advertising. Shockingly, they also set up look-a-like (FB/IG) and act-a-like (Pinterest) audiences, but never actually implemented them into their social media ad targeting. All in all, their account was not intelligently structured to enable a manager to pull key levers that could control budget allocation based on ROAS. There were large volumes of under-performing audience segments, ad copy, and creative that were eating a lot of budget with little return. Immediately, we found huge opportunities for growth in segmenting audiences, ads, and landing pages to increase relevance and conversion while enabling our team to allocate and scale budgets for high-ROAS-performing segments while turning off and cutting low-ROAS-performing segments.

Omni-Channel Segmentation Strategy

In overhauling existing advertising channels and launching new ones, our core strategy centered on granular segmentation of audiences, placements, and keywords so we could closely match ads and landing pages to each segment to make everything as relevant as possible in driving higher membership conversion rates. This segmentation would also allow us to easily see which segments delivered strong ROAS, and therefore, allocate budget to scale or turn off segments based on results. This strategy was applied across all of the following advertising channels:

1. Facebook and Instagram

2. Pinterest 3. Google & Bing Search

4. Display and Youtube

Landing Page Segmentation Strategy

In order to match landing page content with ad copy/creative that would appeal to specific audiences and keyword segments, we decided to create 4 landing page templates with dynamic content insertion systems. Specifically, this meant that we could dynamically change headers, descriptions, images, product category highlights, specific brand highlights, testimonials, FAQs, and other content using URL query string parameters with only 3 landing page templates — as opposed to manually creating 100’s of individual landing pages. For example, we could use the following URL query string parameters for someone interested in jewelry:

- “header=jewelry-box”: “Jewelry and Luxury Goods Subscription Box for Women” - “coverImage=jewelry”: Cover image of a box filled with jewelry products - “productHighlight=jewelry”: Product highlights of jewelry items in stock

We used 3 core landing page templates across all advertising channels:

General Landing Page Template: Used for branded search, competitor search, look-a-like, and remarketing campaigns. Box-Niche Landing Page Template: Used for subscription box search campaigns and niche-audience segments across social, display, and Youtube campaigns — for example, showing a luxury hand-bag specific landing page for the Facebook interest audience of “luxury bags” and for Youtube ad placements on “luxury bag influencer video reviews”. Product-Category-Specific Landing Page Template: Used for category specific search campaigns and niche-audience segments across social, display, and Youtube campaigns — for example, showing a fashion-centric landing page for the Pinterest interest audience of “women’s fashion” and for search queries like “luxury women’s clothing websites”.

Over 16 months, we created over 1,200 unique landing page variants using our 3 landing page templates. Each landing page was crafted to not just attract the visitor but also to succinctly communicate the unique benefits offered in the context of the user's search or audience persona.

This approach was crucial in converting relevant interest into actual membership subscriptions.


Customized Ad Copy and Creative Assets

We designed ad copy and creative assets to resonate with each segment's specific interests and needs — sometimes re-using the clients existing organic social media content and at other times creating our own.

This included aligning the messaging with the savings, product categories, and brands that were most relevant to each audience persona — ensuring a cohesive and personalized user experience from the ad to the landing page.

For example, if the target audience was women in-market for jewelry, the ad copy would highlight unique jewelry and related products they could get at discount by joining through a membership.

We also A/B tested multiple ad formats and variants to find which copy, creative, and formats performed with the highest ROAS across each channel.

Revenue-Centric Tracking Across Channels

The client generated revenue through membership subscription fees and via discounted member purchases of luxury products within their members-only e-commerce platform.

While 83% of their revenue came from member subscriptions, member purchases in the platform still accounted for 17% of annual revenue.

Therefore, for each ad platform, we set three primary conversion events:

Annual membership subscriptions — $345 revenue value:

These subscriptions were paid upfront and indicated guaranteed revenue covering 4 quarterly subscription boxes over 12 months. Given the client offered no refunds, we felt comfortable allocating the full subscription revenue value in determining ROAS of campaigns. Quarterly membership subscriptions — $202 revenue value: While quarterly member subscriptions were priced at $405 annually, four different charges of $101.25 were made each quarter. Given the client had a relatively consistent quarterly churn of 16% on quarterly subscriptions, we calculated 4 quarters of 16% churn in determining the assumed 12-month value of a new quarterly subscription. Member purchases — dynamic $ revenue value: Given advertising platforms limit cookie tracking windows, we could only account for member purchases made within 90 days of a users last ad click. However, this helped us optimize for members who both subscribed and made purchases outside of their quarterly box deliveries.


In order to accurately measure attribution of ROAS across channels, we also implemented a 3rd party tracking tool called Hyros.

Hyros enabled us to see the impact each channel had on converting membership subscriptions across multiple touch-points. For example, if someone converted on a branded Google search campaign, it didn’t necessarily mean that was the only ad the user interacted with. For example, they may have first seen an ad on Pinterest and then looked us up and converted on Google Search.

Omni-Channel Budget and Bidding Optimization

Across channels, we launched new campaigns with initial budgets of $50/day.

Starting conservatively allowed us to gather valuable campaign performance data without excessive spending in blowing monthly budgets without achieving ROAS targets. We set standard Target Cost Per Acquisition (CPA) and Target Return On Ad Spend (ROAS) bidding targets based on historically successful campaigns. After a two-week period of data collection, campaigns showing promising results received increased budgets, while underperforming campaigns were either optimized or paused. This allocation of budget ensured that we maximized the return on investment, continually scaling up the successful campaigns while refining or discontinuing those that did not meet performance criteria.

Facebook, Instagram, and Pinterest Ad Segmentation

In segmenting ads across Facebook, Instagram, and Pinterest, we wanted to achieve a cohesive user experience in making ads and landing pages as relevant as possible to each audience persona while allowing us to have granular control in allocating budget towards the highest performing audiences.

Identifying Qualified Audiences: Looking through the client’s historical advertising data we identified 58 high-performing audiences in addition to outlining over 800+ key audience segments we wanted to test across interests, behaviors, and look-a-like audiences (act-a-like audiences on Pinterest). Clustering Audiences into Ad Sets across 180 Campaigns: Instead of lumping all audiences into one campaign, we launched over 180 separate campaigns with segmented ad sets (i.e. ad groups) that contained 5 to 15 closely related audience segments. This meant that we could iteratively test new audiences to identify which one’s performed and which one’s didn’t — and subsequently increase/reduce budget allocation and bidding targets. Testing Different Ad Formats and Creative: Each campaign tested multiple variants of ad formats and creative that were relevant to each audience segment. For example, we tested fashion-related UGC-styled content, image ads, and video ads for a fashion-centric audience ad set. Once we identified trends in the highest performing ad formats, creative, and copy across our 58 high-performing audience campaigns, we launched new campaigns with similar formats and structures to achieve faster ROAS performance in testing new audiences. On of the biggest insights we gathered was that UGC-style content significantly outperformed other creative on both Facebook and Instagram. Beyond traditional static pin image ads, we found that collection and carousel ads performed the best on Pinterest — these ad formats allowed us to easily highlight the value of accessing a wide variety of discounted luxury women’s products within and across relevant verticals with a membership subscription. Layering Qualified Demographics: All campaigns were layered to only target women between 18 to 54 since we had identified that these segments had produced the highest and most consistent ROAS in the past. Data-Driven Optimization and A/B Testing: After running campaigns for several months, we identified the highest performing audiences, ad copy, and creative and would often break out these targets into their own campaigns so we could isolate budgets on high-performing ROAS segments. For underperforming segments, we would often remove certain audiences, ad copy, and creative all together. We’d then continually A/B test new audience segments and create look-a-like campaigns based on converters, new followers, and people who engaged with content (for example, viewing at least 50% of video content, saving an ad, or sharing). Remarketing to Non-Converters: Throughout our advertising, we made sure to run remarketing campaigns for people who had showed positive engagement with our ads or visited our website without converting.

Google and Bing Search Ad Segmentation

Throughout our partnership, we launched over 140 search campaigns on Google and Bing. These campaigns fit into 4 different categories, all with the primary goal of getting in front of people actively searching for queries, products, and brands that were directly or indirectly relevant to the client’s subscription service:

Subscription-box search campaigns: Campaigns targeting general women’s subscription box queries with separate adgroups for niche-specific subscription box queries that related to what the client offered. For example an adgroup for keywords like “women’s fashion subscription”. Product-category-niche search campaigns: Campaigns targeting different general product and product-category queries. For example, a campaign focused on facial beauty products with different adgroups for mascara, lipstick, foundation, primer, concealer, and other products offered at up to 75% discounted rates for members. These campaigns had highly relevant ads and landing pages that presented the client as the premier subscription for these kinds of products. Competitor campaign: Campaigns targeting people searching for competitors — high-intent audiences that are already aware of and in-market for women’s product subscription services. Branded campaign: Bottom-of-funnel targeting to capture users who searched the client’s brand after seeing ads or hearing about the business from another channel — while ensuring we always showed up in the first position to avoid competitors stealing any traffic that was at a final stage of converting.

One interesting note is that we initially found our search campaigns to be low-performing when we kept campaign targeting at only the keyword level.

Therefore, we decided to shift focus in limiting campaign targets to high-income categories, women between 18 to 54 years old, and the 48 high-confidence account-wide in-market, affinity, and detailed demographic audiences we had seen perform well in the clients historical branded search campaigns.

These campaigns became our “base” campaigns across all the different campaign segments we launched.

When making this shift, we saw a dramatic increase in ROAS given the exclusive limitations helped us better target higher-income individuals who could afford our more premium luxury subscription.

Once we saw a base campaign was working well and achieving a 300% or higher ROAS, we then launched derivatives of these campaigns targeting new audiences, income demographics, and even sometimes testing the campaigns again without any exclusionary criteria. In total, we tested over 300 affinity, in-market, detailed demographic, and custom audience segments.

Once campaigns performed, we also A/B tested different bidding strategies (Target CPA vs. Target ROAS) and different targets in order to find the optimal bidding strategy and targets for each individual campaign.

Most of our campaigns were initially focused on Google. However, in the last 5 months of our partnership, we started launching the most successful Google search campaigns on Bing.

Youtube and Display Ad Segmentation

Beyond traditional remarketing for non-converting website visitors, we tested ~90 layered campaign strategies across keyword content placements, domain and webpage placements, channel and video placements, and YouTube search campaigns.

All campaign strategies we’re initially layered to exclusively target a range of 33 in-market audience segments we had identified as high-converters from search campaigns. This ensured we only reached people who Google classified as in-market for the different product categories the client’s subscription offered.

We also layered all campaigns to only target women aged 18 to 54. Initially, we also layered targeting to only show ads for women in the top 10% to 30% household income categories. However, we eventually expanded our base campaigns to test variant campaigns of additional household income categories and audiences (starting on low budgets).

The goal was to identify campaign strategies, keyword content, audiences, demographics, and placements that drove profitable subscription conversion rates — showing up to these qualified audiences the moment they were consuming content that related to products offered through our subscription service and tailoring unique ads and landing pages to each segment.

Keyword Content Placements: We started by implemented a bottom-of-funnel strategy by placing ads on webpages and YouTube videos that specifically mentioned keywords that had proved to have strong conversion rates in search campaigns — these keywords included subscription box keywords, product category keywords, and brand product keywords. This process involved creating segmented lists of 20 to 100 keywords and phrases, ensuring that the ads appeared in contexts directly relevant to the target audience. These segmented lists were then clustered into adgroups within individual campaigns that had a consistent theme (i.e. “subscription box keywords”, versus “luxury fashion keywords” vs “skincare keywords”, etc). Domain and Webpage Placements: By analyzing data from keyword content placements, we identified domains and pages with the best performance in terms of engagement and relevance. New campaigns were then created (focused on these domains), targeting pages that had good engagement and specifically mentioned high-converting keywords. This targeted approach helped to reach a more relevant and interested audience, enhancing the efficiency of ad spend. YouTube Channel and Video Placements: Recognizing the effectiveness of targeted YouTube placements, we also researched and segmented relevant YouTube channels and videos for ad placements across each product vertical. Namely, we targeted channels and videos from top brands and influencers with content directly related to the products our client offered. By selecting these specific channels and videos, they ensured that messaging was delivered to an engaged and targeted audience, increasing the potential for capturing the interest of prospective members. YouTube Search Targeting: We also launched YouTube search campaigns targeting high-performing keywords we had identified in traditional search campaigns.

The Results

102k+ New Member Subscriptions

Over the 18-month period, we were able to acquire over 102,000 new member subscriptions at an average acquisition cost (CPA) of $107.15 per new member.

In total, we acquired 44k+ new annual member subscriptions (43% of new members) and over 58k new quarterly member subscriptions (57% of new members).

3.4x Increase in Monthly Ad Spend

While our initial goal was to 2x their ad spend to $320,000 within 12 months while retaining a 250%+ ROAS, we managed to achieve $320,000 in monthly ad spend by month 9.

By month 18, we achieved a remarkable $705,000 in monthly ad spend with above 250% ROAS — a 3.9x increase versus the first month and a 3.4x increase when averaging monthly spend across our 18-month partnership ($610,000 average monthly ad spend).

$29.4M In New Client Revenue with 268% ROAS

With an average revenue per member (ARPU) of $287.16, that meant we returned $2.68 for every dollar spent on advertising — generating over $29.4M in additional revenue from advertising for the client.

In only being able to account for member purchases made within the first 90 days, subscription revenue accounted for 92% of revenue attributed to our advertising efforts — roughly $15.2M came from new annual member subscriptions with $11.8M coming from new quarterly member subscriptions ($27M total).

With around 38% of new members making between 1 to 4 purchases within 90 days of their last ad click (separate from their subscription), we were able to track revenue for over 97,300 member purchases with $24.91 average commission revenue per purchase. These purchases accounted for roughly 8% of revenue attributed to advertising (~$2.4M).

We were surprised at the ROAS we were able to achieve, especially considering these numbers did not include any additional revenue from annual renewals, referrals made from each new member, or member purchases made after 90 days.

These numbers also included conservative 12-month assumed revenue from new quarterly subscriptions by accounting for 16% average quarterly churn on these subscriptions.

In reality, our 12-month LTV ROAS was likely a lot higher than 268%.

$6.9M Operating Profit from Ads with a 23.7% Margin

While we don’t have data on the clients overhead and administrative costs, we do know they operate at a 61% gross profit margin (profit on cost-of-goods-sold) and could use this to calculate operating profit from ads (profit on COGS and ad spend).

When factoring in all ad expenditure and a 39% average cost-of-goods-sold (COGS), we determined that the client was able to retain roughly a 23.7% operating profit margin on it’s advertising operations with all costs included.

With ~$11.4M in cost-of-goods-sold (COGS) and ~$10.98M in ad expenditure (over 18 months), that meant that the client was able to retain ~$6.9M in operating profit from ads through our partnership.