B2B Ecommerce Marketing Strategies: How Dynamic Drop-Off Remarketing, LinkedIn Ads, and Personalized Email Campaigns Grew Ad Revenue by 3.1x for a Custom Swag Store

Case Study Summary

The Challenge

The Solution

The Results

The Challenge

Our client, a leading online B2B e-commerce product manufacturing retailer, offers custom swag, merchandise, and other products to over 200,000 satisfied B2B and B2C customers with partnerships across several top Fortune 500s.

While internal teams had managed to successfully scale up their paid advertising efforts to ~$115,000 per month, the client felt that their growth trajectory was hitting a ceiling.

Launching LinkedIn as a New Advertising Channel

Given their heavy B2B focus in selling corporate swag, company promotional merchandise, and other items, they always felt they had a large untapped opportunity in LinkedIn.

After being referred to us by an existing client, they decided to take us on for piloting LinkedIn advertising as a new channel.

Our challenge was to identify how to iteratively launch LinkedIn advertising with 200% to 300% ROAS on initial order values in the first 3 to 4 months before scaling further.

Fixing the Leaky Bucket — Cart Abandoners and Non-Converters

As part of piloting our partnership, the client also wanted us to identify opportunities to improve the ROI of their existing channels — Google search, PMax shopping, and static remarketing campaigns on social and video/display.

With a staggering 86% cart abandonment rate across channels, they primarily wanted us to identify how we could “fix the leaky bucket” in getting more visitors to convert — or in the way that we saw it, to get them to come back and convert.

Improving Initial Order ROAS and 6-Month LTV ROAS

Given the client serviced business of all sizes (with some focus on B2C), their average order values (AOV) varied widely from $100 up to $10,000+ — with an average AOV of $482.

In looking at their advertising accounts, we saw that they had achieved an average Return-On-Ad-Spend (ROAS) of 284% ROAS on initial order values for the preceding 12-months.

When looking at internal data, we saw that 6% of new customers made new purchases (1.1x purchase frequency at a $512 average AOV) within 6-months of their initial order — 304% 6-month LTV (Life-Time-Value) ROAS.

We were tasked with identifying opportunities for improving both metrics, with a priority focus on improving 6-month LTV ROAS.

The Solution

When diving into the challenge of launching a new advertising channel with LinkedIn and optimizing immediate and 6-month LTV ROAS for existing channels, we took creative approaches to identify gaps for how we could improve customer conversion while growing ad spend, top-line revenue, and bottom-line profitability.

Here’s how we tackled it:

Identifying Reasons for 86% Cart Abandonment

The key insight we gathered from looking at analytics on purchase behavior was that orders often had a 2 to 8+ week lead time with 3 to 10+ unique sessions from initial visit to initial purchase.

Therefore, the 86% cart abandonment was not truly indicative of ~9 out of 10 visitors not making a purchase. It just meant there was 86% cart abandonment on individual sessions.

Through customer interviews and conversations with their team, we identified the following reasons for longer lead times:

1. Minimum order quantities (MOQ) with high average order values (AOV) meant longer consideration time before making initial purchases

2. Many first-time visitors were in research or consideration phases

3. Some buyers needed budget approval before making orders

4. All buyers needed to design and customize their merch before ordering — requiring multiple website visits in selecting their products, making their designs, and then uploading their designs for final purchase

5. Many buyers were testing multiple products and designs to review with their team before deciding on which purchase they would make

Given the multiple touch-points needed to convert customers, we felt remarketing and email optimization would play the biggest role in converting more customers.

The client had already invested significant resources in optimizing their store UX/UI for conversion, therefore we felt that the larger opportunities laid in the advertising itself.

Collecting Email Upfront for Promotional Offer

Given the 86% session cart abandonment rate, the one opportunity we did identify for optimizing their store UX/UI was to incentivize new visitors to give us their email upfront by offering them 20% off their first order.

We had their team place optional email collection inputs for 20% off in the first step of ordering, in their navigation menu, and on every product page.

With 81% of new customers inputting their email as the first step in ordering, this allowed us to nurture new leads using first-party data (emails) in addition to using 3rd party data through remarketing cookies.

We used these emails to run personalized email campaigns in addition to email-based remarketing and look-a-like audience campaigns.

This step proved critical in enabling us to engage customers from their initial touch-points all the way through their initial purchase.

In their existing setup, the client had left email collection for the last step of purchasing.

Faster Signals via Staged-Order Conversion Tracking

Given the 2 to 8+ week lead time in converting new customers, we knew we had to identify ways of getting faster conversion signals into our advertising in order to allow both the AI-bidding algorithms and our team to have more real-time data readily available to optimize campaigns — both in existing channels and before launching LinkedIn.

Issues with Current Conversion Tracking Setup: Before hiring us, the client had implemented standard e-commerce conversion tracking across Google and Meta. Their tracking was purely centered on purchases — using e-commerce data layers to dynamically track revenue values from completed orders. While we would normally suggest this setup for standard e-commerce companies, the problem with this setup was that conversion signals could often take 2 to 8+ weeks to show up in advertising campaigns. Therefore, the campaigns were very difficult to optimize given the long delay between ad clicks and purchase conversion signals. It’s also important to note that purchases made 90 days after a user’s last ad click (for example, from email remarketing campaigns), wouldn’t register any value in our advertising given cookie policy tracking-windows. We also hypothesized that this conversion setup optimized campaigns for smaller AOVs given smaller purchases required less consideration and time than larger ones — we got this hunch by comparing their organic initial order AOVs ($547 AOV) to their paid initial order AOVs ($482 AOV). Staged-Order Conversion Tracking — 4 Primary Conversions Used in Bidding: In order to fire faster signals into our advertising, we decided to set primary conversions for add-to-cart and checkout lead stages (with staged-out proxy conversion values) in addition to tracking purchases and sample orders with real revenue conversion values:

1. Promotional email input — $25 proxy value Proxy value of 25% of minimum order values ($100)
   

2. Complete design and order details — $50 proxy value Proxy value of 50% of minimum order values ($100)
   

3. Sample order — dynamic $ revenue value Actual sample order revenue value
   

4. Complete purchase — dynamic $ revenue value Actual order revenue value minus $75 proxy values (assigned in first 2 steps)

While this setup over-estimated ROAS within our advertising accounts — given many people would input their email or complete design/order details without ever making a purchase — it did enable us to optimize campaigns using real-time data (see next section). In determining the actual ROAS and profitability of our advertising, we would simply reference the values of actual orders versus spend to ensure we were achieving our real ROAS targets. Ability to Optimize Campaigns Real-Time: With 6.92% of ad clicks converting on our “promotional email input” conversion goal, this proved to be a game-changer in our ability to optimize campaigns. This conversion gave us a high-volume of real-time signals that enabled us to quickly assess the success of new campaigns, targeting, and experiments — without having to wait 2 to 8+ weeks to see whether the campaign led to purchases at our target purchase ROAS. Even though most new campaigns would show a loss on purchase ROAS in their first 2 weeks, we found that if a campaign was able to achieve a CPA of under $45 for promotional email inputs (real-time signals), than it was likely going to perform well in terms of purchase ROAS after the 2 to 8+ week lead time window had passed. While our “complete design and order details” conversion signal still often had a longer delay, it added an extra layer of qualified conversion tracking that allowed us to assess the performance of campaigns much quicker then with having to wait for the actual purchase to come through.

Dynamic and Staged-Order Drop-Off Remarketing

In identifying that the client’s customers took 3 to 10+ sessions on average to convert, we knew that an intelligent remarketing strategy would be one of the biggest drivers of profitable growth for the client — second only to email marketing campaigns.

Throughout the 16-month partnership, we found that remarketing campaigns accounted for a 162% uptick in account-wide conversion rates from ad traffic.

Dynamic Remarketing on Google, Youtube, Display, and Facebook: While the client had been using static remarketing for website visitors across Google, Facebook, Instagram, YouTube, and display, we actually discontinued static remarketing across these platforms. We instead replaced static remarketing campaigns on these platforms with a much more sophisticated method that showed users the actual products they were looking to buy — dynamic product remarketing. Dynamic remarketing ads were tailored to individual user behavior and displayed ads, showcasing the exact products that users had viewed on the website. Facebook, Instagram, and Google allowed for Dynamic Products Ads (DPA’s) that automatically integrated products that the user had shown interest in on the site in the advertisement itself. These highly personalized ads served as powerful reminders and encouraged users to return to the site to complete their purchase. Given we couldn’t do dynamic product remarketing on LinkedIn, we instead used static remarketing to target website visitors on LinkedIn who had not completed any of our conversion actions. Staged-Order Drop-Off Remarketing Across All Channels: With our staged-order conversion tracking setup complete, we then created audience segments based on which stage of the buying journey people had completed — indicated by our primary conversion signals. We created these audiences by using both cookie-data for conversion events in addition to integrating offline first-party data (emails) through conversion APIs. With the eminent discontinuation of cookie-tracking, this ensured our setup was future proof while enabling us to capture people who had blocked cookie tracking. We then layered these audience segments into our dynamic remarketing campaigns across Google, Youtube, Display, and Facebook in order to couple stage-of-the-buying-journey-specific copy with the actual products people were looking to purchase. For LinkedIn, we simply used these audience segments to run more effective remarketing campaigns referencing personalized copy and CTAs that were relevant to each user’s stage of the buying journey. For all stages, we A/B tested different retargeting ad copy, CTAs, and assets. For people who had only filled out our promotional email input, example copy included:

• “Last step — Upload your designs for 20% off your first order”

• "Need help with your design? Chat with a design expert”

For people who had completed their designs and order details, examples included:

• “Your order is ready to be delivered”

• “Finish your order today for 20% off”

Iterative Approach to LinkedIn Advertising

In setting up an iterative entry into LinkedIn Advertising, our goal was to start with the highest-confidence segments that could deliver faster ROAS and give the client confidence to continue increasing their investment into LinkedIn.

Initial LinkedIn Campaigns: Given we had no benchmarks on what to expect for cost per result, we set each campaign to run on maximum delivery (bidding strategy) and started by only using in-feed ad placements — typically the highest performing within LinkedIn. For each campaign, we included 10 to 20 variances of ad copy, assets, and CTAs to see which ones performed the best. The first campaigns we launched included:

1. Remarketing campaigns A blended mix of staged-order drop-off remarketing and static remarketing for people who hadn’t completed any conversion goals. These campaigns quickly established strong ROAS performance given they were re-engaging customers who had already visited our site from other paid and organic channels — as opposed to prospecting for new customers who had never heard of us.
   

2. Look-a-Like Audience Campaigns Based on Customer Lists In starting our first prospective campaigns, we wanted to reach new people who shared similar profiles to our existing customer base — relying on LinkedIn to do the initial work for us in identifying which industries, company sizes, job titles, and other LinkedIn targeting criteria proved most effective. This allowed us to limit the amount of initial campaigns we launched while keeping initial budgets low so we could learn which LinkedIn targeting criteria proved most effective before launching separate campaigns targeting these segments explicitly.

Special Note: LinkedIn Look-a-Like Audience Discontinuation With LinkedIn discontinuing the ability to use Look-a-Like Audiences as of February 29th, 2024, we switched over all Look-a-Like Audience Campaigns to use predictive audiences and audience expansion. Look-a-Like Audience Campaign Segmentation: Using over 175,000 emails the client had from prior customer purchases, we segmented customer lists for look-a-like audiences by two criteria:

1. Product Category Order Look-a-Likes Creating separate look-a-like audiences for customers who had purchased different product categories. For a example, a list of customers who purchased custom t-shirts and a separate list for customers who purchased custom water bottles. The goal here was to identify whether we could segment key LinkedIn targeting criteria for each product segment to make future prospecting campaigns more relevant with higher conversion for each segment. We initially launched five campaigns based on the highest performing product categories.
   

2. Average Order Value Look-a-Likes We created four separate customer lists based on historical average order values: one list for low AOVs (under $300), another for average AOVs (between $301 to $600), a third for upper-tier AOVs (between $600 to $1,000), and a fourth for high AOVs (above $1000).
   

Insights from Performance of Initial Campaigns: Starting with a budget of $10,000 per month, we quickly found that LinkedIn was a profitable channel. By the end of the second month, we were already achieving consistent CPAs under $45 for promotional email inputs (highly indicative of profitable future purchasing) and had actually returned 193% ROAS on orders made within the initial 60-day period of launching advertising on LinkedIn — outperforming several of the clients Google and Facebook campaigns. In looking back at the performance data after 8 weeks (accounting for orders that came with longer lead times), we found that our first two months advertising on LinkedIn had generated over $61,000 in revenue from initial orders on $20,000 in spend (305% ROAS). Given most of our campaigns were limited by budget, the client approved us tripling LinkedIn Advertising budget to $30,000 per month in the third month. Scaling LinkedIn Advertising Campaigns: After seeing promising initial results, our challenge now came in scaling our ad spend while maintaining or improving ROAS. To do this, we started by further segmenting our look-a-like audience campaigns to be able to scale budgets across segments based on CPA/ROAS performance — giving us granular budgetary control on each segment. We then created new prospective campaigns based on high-performing LinkedIn targeting criteria we saw within our look-a-like and remarketing campaigns. These were new campaigns that no longer relied on look-a-like audiences. For example, we identified that job titles related to “recruiter/talent acquisition” and “manager” often had the best performance within medium to larger companies (200 to 10,000+ employees) while job titles like “Founder/President/CEO” often had high conversion rates among smaller companies (11 to 200 employees). We also identified similar trends for high-performing segments based on industries, demographics, job experience, education, interests, and traits. Based on these insights, we broke out over 40 hyper-granular campaigns with layered targeting that tested combinations of all of the above segments based on which criteria performed well in remarketing and look-a-like audience campaigns. Over 16 months, we were able scale the clients advertising budget on LinkedIn from $10,000 to over $90,000 per month. Optimizing LinkedIn Advertising Campaigns: In optimizing our LinkedIn campaigns, we primarily looked at which segments, targeting, ad copy, and creative had high performance versus which ones didn’t. The core approach was to isolate and add budget into high performers while cutting or adjusting low performing segments. For high performers, we’d break them out into their own campaigns to be able to allocate more budget into segments that delivered strong ROAS. For low performers, we would cut segments, ad copy, and creative that didn’t drive results or sometimes test different copy, creative, and audience layer combinations to see if we could improve performance before cutting. With bidding strategies, we primarily relied on maximum delivery. However, we started testing manual bidding strategies for landing page clicks to try and better optimize our CPAs for website conversions — while this worked on some campaigns, we found that maximum delivery often worked best. Across campaigns, we tried multiple ad formats and found that image, video, and carousel ads consistently outperformed other formats when delivered in-feed (sponsored ads). While we tested new campaigns for spotlight (dynamic) and sponsored messaging, we found that relying on sponsored content in-feed always delivered the best results.

Personalized Email Automation Using Bloomreach

Although the internal team managed email campaigns using Bloomreach, we did get involved with strategizing ways to improve open rates, click-through rates, purchase conversion, average order values (AOV), and repeat purchases (LTV):

Primarily, we centered our involvement on:

1. Experimenting to improve key metrics

2. Personalizing campaigns to convert promotional email inputs

Experimenting to Improve Email Marketing Metrics: Our experiments centered on creating more narrow segmentation of email lists based on the stage of the buying journey, product category interest, past purchase behavior (if available), geography, and email engagement rate. Across this segmentation, we collaborated with the team to experiment across:

• Headlines and copy (focusing on short and sweet)

• Email frequency and timing of delivery

• HTML emails (designed) versus standard text (text-only)

• Company-as-sender versus individual-as-sender

• CTAs, discount, and urgency messaging

• Dedicated landing pages

• Personalization of email components

• Audience-specific email flows (sequences)

We also helped prune email lists so we could avoid high unsubscribe rates while keeping open, click-through, and conversion rates higher to ensure we maximized inbox delivery across segments. Personalized Email Campaigns to Convert Promotional Email Inputs: To make the most of the new conversion signal we introduced (promotional email inputs), our experiments primarily consisted of strategizing ways to maximize the ROI and effectiveness of converting these emails into first-time purchases. For example, for visitors who entered their email into a product-page promotional input without uploading a design, we found strong results in sending automated HTML5 emails — from a standard notifications email — 30 minutes after they closed their session. These were polished email designs with product images and a CTA button to “Finish Your Design”. These campaigns garnered 35.1% open rates (vs average of 19%), 3.9% click-through rates (vs average of 1.4%), and 1.42% conversion rates on purchases (vs average of 0.77%). We also found strong results in sending follow-up email flows from an actual team member email (with name and photo), 2 hours after their session drop off. These were text-only emails that felt as if the person had actually taken the time to reach out directly with a CTA to help them in polishing their design and order details. The campaigns garnered 28.4% open rates, 2.6% click-through rates, and 1.15% conversion rates on purchases.

Up/Cross Sell Remarketing and Email Campaigns (Post-Purchase)

Once we had converted customers on initial orders, we focused audience segments to cross and upsell customers via post-purchase remarketing and email campaigns.

We used dynamic product remarketing ads (DPA) to cross/upsell relevant products across Google, Display, Youtube, Facebook, and Instagram to both existing and new customers (starting 60 to 90 days after making their first purchase). For example, remarketing branded sunglasses to people who had purchased custom t-shirts.

Across LinkedIn and all other advertising channels, we also did cross/upselling for repeat product orders based on occasions. For example, for the end of the year, Black Friday, Christmas, and other seasonal occasions. For these campaigns we used promotional discounts and personalized messaging for repeat customers.

We leveraged the same strategies in post-purchase email campaigns to re-engage each of our audience segments based on their email engagement rate, order history, and product category interests.

As an additional growth layer, we also sent out “referral” email campaigns incentivizing current customers to refer colleagues or friends to make their first order. While we didn’t attribute revenue from these campaigns to our efforts, referrals did play a big part in growing the client’s new customer acquisition.

Additional Google Ads Campaign Optimizations

While most of our efforts over the 16-month period were focused on revamping the client’s remarketing efforts, we did also make significant optimizations across their search and Performance Max (PMax) shopping campaigns.

Performance Max (PMax) Shopping Optimization: Initially, we focused on breaking out their PMax shopping campaigns to have a more narrow and focused set of products in each campaign. We clustered listing groups (products) into each campaign based on 5 to 10 products that shared similar ROAS performance. This allowed us to set more accurate ROAS targets at the campaign-level while allocating budget accordingly based on performance. By the end of it, we broke out the client’s 8 PMax campaigns into 17 separate campaigns while launching 4 new PMax campaigns for products they had not yet advertised — all with their own bidding targets. We also expanded the client’s use of assets, audience signals, and search themes in order to give the campaigns more to work with while at the same time adding in negative keyword lists to better refine performance. Once we saw our expanded targeting and segmentation was working, we continually refined campaigns by running A/B experiments for different bidding strategies (Target CPA and Target ROAS) with different bidding targets to maximize net monthly ad profit. Search Campaign Optimization: Since Google still prioritizes search campaigns over PMax, we also expanded their use of traditional search campaigns. We first started by optimizing their existing campaigns through:

• Keyword and match-type optimization

• Search terms analysis and negative keyword lists

• Improved ad copy relevance and quality score

• Expanded use of assets/extensions

• Improved landing page relevance

• Modified bidding strategy targets

We launched our changes on A/B experiments so we could easily prove better performance before applying our changes to replace existing campaigns. With more and more data on which keywords and campaigns delivered the highest ROAS, we then launched 11 new campaigns across product/category keyword segments, a competitor campaign, and a branded campaign.

170% Increase in Google Ad Spend: When the client first hired us, they were spending about $90,000 on Google search, PMax, Display, and Youtube campaigns. By the 16th month of our partnership, we had expanded their remarketing, PMax, and search efforts to over $150,000 in monthly spend with 117% increase in account-wide ROAS.

Piloting Amazon Ads for B2C Non-Branded Items

Throughout our partnership, the client had started testing an exclusively B2C focused play for selling popular merchandise they had designed for direct sales on Amazon.

Given Amazon would charge between 8% to 17% referral fees in addition to closing and fulfillment fees on orders, the client had historically avoided selling anything on Amazon — especially given the majority of their sales and profit came from B2B bulk orders with minimum order quantities (MOQ).

However, they felt an additional opportunity lied in producing a large quantity of B2C merch (designed in-house and made in bulk) that avoided the higher costs of smaller-batch custom manufacturing from designs uploaded by each customer on their site.

The goal of this new vertical was to see if their in-house design team could capitalize on market trends in producing popular B2C items at scale to be sold on Amazon.

We collaborated with another agency (focused on organic optimization of Amazon product pages and stores) to start running ads that would drive more traffic and reviews to these items.

So far, we’ve run ~$15,000 per month on the most popular items that scaled organically with a slight profit margin. However, we feel confident that with more data coming in we should be able to scale Amazon B2C as a new growth channel in the coming months.

The Results

Our unique approach to converting a higher percentage of ad traffic on existing channels in addition to scaling LinkedIn as a new advertising channel, yielded remarkable results over the first 16 months of our partnership.

Here's a detailed overview of the outcomes we achieved:

257% Growth in Monthly Order Volume

Before our partnership, the client was getting roughly 675 new orders per month at an average cost-of-acquisition (CPA) of ~$169 — generating roughly $325k in monthly ad revenue from initial orders on $115k in monthly ad spend.

In averaging performance across our 16-month partnership, we were able to:

• More than 2x ad spend to over $260k per month

• Increase initial order value ROAS from 284% to 347%

• Reduce CPA per new order by 12% from ~$169 to ~$149

The combination of the above factors increased the client’s average monthly order volume to over 1,700 new orders per month — generating over $900K in monthly ad revenue.

While all of our efforts contributed in helping us to accomplish this, the most impactful contributors were the launch of LinkedIn as a new advertising channel in addition to the complete revamp of the client’s remarketing strategy and conversion tracking setup.

3.1x Growth in Monthly LTV — $17.3M in 6-Month LTV Revenue

Beyond increasing the client immediate ROAS from initial order values, our main KPI was focused on increasing 6-month LTV ROAS through higher ad spend, more effective remarketing and email campaigns, and improved ad optimization centered on getting repeat orders with higher average order values (AOVs).

As a benchmark, the client was used to getting roughly 6% repeat purchasers from new customers within 6-months of their initial order.

These customers had an average repeat purchase frequency of 1.1x (i.e. most repeat purchasers only completed 1 additional order within 6 months), with higher AOVs ($512 AOV) then initial orders ($482 AOV).

Through email campaigns and up/cross sell remarketing, we were able to more than double the repeat purchase rate among new customers to 14% while increasing repeat purchase frequency to 1.3x — with higher repeat purchase AOVs of $565 in addition to higher initial purchase AOVs of $518.

The above in combination with increased ad spend and optimization across channels, increased average monthly LTV revenue from $350k to ~$1.1M per month — generating over $17.3M in 6-month LTV revenue across our 16-month partnership.

4.5x Increased Operating Profit from Ads — from 19.1% to 27.95%

As a symptom of increased ad spend in combination with drastic improvements on both immediate and 6-month LTV ROAS, the client experienced a sharp spike in monthly operating (OPEX) profit from advertising.

When factoring in ad spend and the client’s 48% cost-of-goods-sold (COGS) — 52% gross profit margin — we managed to grow the client’s “monthly operating profit” and “operating profit margins” on advertising as follows:

Based on Initial Order Revenue: 3.9x Increase:

• Before: $54k monthly profit | 16.79% margin ($325k initial order revenue - $156k COGS - $115k ad spend)
  

• After: $209k monthly profit | 23.18% margin ($902k initial order revenue - $433k COGS - $260k ad spend)
  

Based on 6-Month LTV Order Revenue: 4.5x Increase:

• Before: $67k monthly profit | 19.1% margin ($350k 6-month LTV revenue - $168k COGS - $115k ad spend)
  

• After: $302k monthly profit | 27.95% margin ($1.081M 6-month LTV revenue - $519k COGS - $260k ad spend)

The significant increase in monthly operating profit from advertising is now enabling the client to continue to grow their investment in advertising and new verticals for their business.

Over 100k New Email Subscribers

Over our 16-month partnership, we also managed to collect over 100,00 additional email subscribers through our promotional email input on their site.

While remarketing campaigns led to a 162% uplift in conversion rates, personalized email marketing on promotional email inputs increased our conversion rate by 311% — from 1.18% to 3.68%.