Social media creators have changed how customers discover products.
A creator wears a pair of earrings. A beauty product starts showing up across TikTok. A nostalgic collaboration catches someone’s attention. Suddenly, a customer who wasn’t even in shopping mode is considering a purchase. The creator may spark the interest, but the brand still owns what happens when that customer has a question, makes a purchase, or needs help afterward.
That creates an enormous opportunity for brands. But generating interest is only the beginning of the customer journey.
What happens when that interested customer has a question?
To find out, we mystery-shopped four brands using social media creators to drive product discovery and demand across multiple stages of the customer journey.
We started where creator-driven product discovery happens: on social media, engaging directly with influencer- and brand-posted content as prospective customers. From there, we followed the journey through pre-purchase support, purchase and fulfillment, post-purchase support, and returns.
We also tested how each brand responded when a specific concern stood between a prospective customer and a purchase.
The biggest finding: generating purchase intent doesn’t mean your CX operation is prepared to capture it.
An unanswered product question, a difficult path into support, outdated automation, or a poorly handled exception may seem like an isolated CX issue. But when a customer has already moved from scrolling to considering a purchase, those gaps can interrupt the momentum creators helped build.
The Brands We Shopped
Mass-Retail Clean Beauty Brand
A clean beauty brand with broad retail distribution and products spanning skincare, makeup, and personal care
Prestige-Portfolio Beauty Brand
A prestige beauty brand known for premium makeup and skincare sold across its own and major retail channels
Accessible-Luxury Jewelry Brand
A digital-first jewelry brand offering accessible luxury through both traditional purchases and a paid membership model
Legacy Collectible Jewelry Brand
An established jewelry brand built around collectible designs, nostalgia, and high-profile collaborations
What We Measured
To evaluate how brands handle demand generated through social media creators, we tested the customer journey from initial product discovery through purchase, support, and returns. We engaged with influencer- and brand-posted content, asked pre-purchase questions, made purchases, contacted support with real post-purchase issues, and completed the return process.
We also tested how each brand handled a specific concern standing between a prospective customer and a purchase. Across the journey, we measured how effectively each brand responded to customer intent, made support accessible, provided useful information, handled exceptions, and helped customers move forward.
Test Type
What We Did
What We Measured
Social Discovery + Engagement
Engaged with influencer- and brand-posted content using comments and product questions a prospective customer might ask
Brand and creator response, response time, usefulness of engagement, whether purchase questions were addressed
Pre-Purchase Support
Asked product questions through direct social and owned support channels
Ease of access, response time, product knowledge, completeness, personalization, and support for purchase decisions
Purchase + Fulfillment
Purchased products and tracked the experience through delivery
Checkout friction, order communication, shipping and fulfillment, packaging, and product condition
Post-Purchase Support + Returns
Contacted support with real post-purchase issues and completed the return process
Ease of reaching support, self-service, response and resolution, handoffs, return experience, and follow-through
Purchase-Barrier Support
Approached each brand separately with a specific concern standing between us and a purchase
Whether the concern was addressed, usefulness and specificity of the response, and whether the interaction helped remove the purchase barrier
5 CX Signals That Stood Out
1. Social Commerce Is Creating Conversations That Customer Support Isn’t Joining
Social platforms have become powerful engines for product discovery. They also give customers what appears to be a direct line to the brands they’re discovering.
We tested that connection by engaging directly with influencer- and brand-posted content, leaving the kinds of comments and product questions a prospective customer might ask while discovering a product through social.
Results were mixed. Some interactions generated engagement, while others received no response from either the creator or the brand. But they raise an important CX question: when influencer content creates purchase interest in the comments, who owns what happens next?
We contacted all four brands through TikTok with pre-purchase questions about products we were considering buying.
None responded.
One brand eventually saw and even accepted our message, but never answered it. Another indicated that it typically replied within minutes, but our question still went unanswered.
That’s a significant disconnect for brands benefiting from social commerce.
A customer who discovers a product through social media doesn't necessarily separate marketing from customer support. They see the brand account, find the message button, and ask the question standing between them and a purchase.
Our later social tests showed that brands approached customer conversations differently across platforms. Some engaged with prospective customers on Facebook or Instagram, while one brand chose not to accept Facebook messages at all.
Choosing not to accept messages on a channel might sound like a drawback — but if you don’t support the channel, blocking messages there creates a clearer customer experience than appearing available and leaving questions unanswered.
For brands generating demand via social media creators, the social media customer experience doesn’t end when a user discovers the product. The comments, questions, and messages that follow are a critical part of the buying journey.
2. Sometimes the First Customer Support Problem Is Reaching Customer Support
In several cases, we struggled to connect with the help we needed.
One brand’s self-service return portal was broken.
Another brand’s email link added friction by launching a mail application we didn’t use instead of opening a contact form.
The most surprising friction came from a brand that required double opt-in for customer support requests.
After submitting the form, we had to open a verification email and confirm our email address before the request ever reached customer support. A small notice appeared after submission explaining the additional step, but it wasn’t prominent enough to catch our attention. And if we took too long to confirm our email address? We’d get an error that the link was no longer available. We missed the required step despite actively looking for friction and unusual experiences throughout the mystery shop.
Every additional step creates another opportunity for customers to drop off.
3. Useful Automation Requires Ongoing Ownership
AI and automation played a significant role throughout this mystery shop, and some of the automated experiences were among the most useful interactions we had.
When we asked the prestige-portfolio beauty brand how one of its blushes compared with the competitive product we normally used, its virtual support experience provided a detailed comparison while acknowledging what it couldn’t confirm from the available product information.
Later, we challenged the same brand about contradictory language regarding whether all of its products were vegan. The virtual assistant engaged with multiple follow-up questions, acknowledged the inconsistency, disclosed that it was a virtual support assistant when asked, and eventually drew a clear boundary around what it could and couldn’t independently verify.
The accessible-luxury jewelry brand’s AI also handled a significant purchase barrier directly. After we raised concerns about reports that shoppers were unknowingly enrolling in a paid membership, the assistant explained the difference between guest pricing and membership pricing and how to purchase without joining.
Automation wasn’t universally successful, though.
One automated support response included self-service links left over from an earlier version of the company’s ecommerce model. The business had changed how customers purchased its products, but the automation hadn’t kept pace. The links led to 404 pages.
Launching automation isn’t the finish line.
Knowledge changes. Policies change. Websites change. Products change. Business models change. The systems answering customers have to change with them.
AI operations need ongoing ownership of the knowledge, workflows, links, escalation rules, and exceptions underneath the customer-facing experience. Otherwise, an automation designed to reduce friction can eventually start creating it.
4. Simple Journeys Test Your Workflows. Exceptions Test Your Operation.
Some of the routine post-purchase experiences in our mystery shop were remarkably easy.
The prestige beauty brand provided a straightforward self-service return process and shipping label. Returning the mass-retail beauty products through the retailer was also simple.
The bigger operational test came when the situation stopped following the expected path.
Our legacy collectible jewelry brand order arrived with damaged packaging and a product that appeared damaged. The return portal also failed when we attempted to use it.
We explained the problem and explicitly requested a refund rather than a replacement.
The warranty process subsequently offered us a gift card or replacement.
We again requested a refund to the original payment method. When we didn’t receive a response, we followed up.
We were eventually told that a $47.84 refund had been issued, even though the order totaled $53.82. That required another follow-up before the company confirmed a full refund.
None of those individual steps represents an extraordinary customer request. Together, they reveal how easily context can disappear as an interaction moves through different workflows.
The customer had already stated the desired resolution. The company had the order information. Yet the customer still had to repeatedly move the issue toward the requested outcome.
Self-service made some of the common interactions in our mystery shop remarkably efficient. But customers don’t always follow the common path. Products arrive damaged. Charges look unfamiliar. Policies create confusion. Customers ask questions the knowledge base didn’t anticipate.
Companies need clear ownership, decision authority, escalation paths, and access to customer context when those exceptions emerge.
5. Answering a Customer’s Question Isn’t Always the Same as Removing the Purchase Barrier
For our final test, we approached each brand as a prospective customer with a genuine reason we were hesitant to buy.
We asked one jewelry brand about allegations that a product design had been copied from another creator. We asked another whether customers could purchase without accidentally enrolling in a paid membership. We challenged contradictory language about vegan products at a beauty brand.
And we told another beauty company that reports of skin reactions made us nervous about trying its products on extremely sensitive skin.
We wanted to know whether each brand’s response helped us make the purchase decision.
The differences were revealing.
The mass-retail beauty brand engaged directly with the sensitive-skin concern on Instagram. It recommended specific fragrance-free products, pointed us toward options accepted by the National Eczema Association, and suggested patch testing and introducing products individually.
The accessible-luxury jewelry brand’s AI addressed the membership concern directly and explained how to purchase as a guest.
The prestige-portfolio beauty brand’s virtual assistant stayed engaged through increasingly difficult questions about its vegan claims. It couldn’t independently resolve the contradiction on the website, but it clearly stated the limits of what it could verify rather than inventing certainty.
The legacy collectible jewelry brand acknowledged our concern about the design allegations and told us the feedback had been passed to the appropriate team. But it didn’t answer the question preventing us from purchasing the product.
That’s the difference between answering a question and understanding customer intent.
A response can still leave the customer exactly where they started: interested in buying, but without enough confidence to do it.
Customer support can remove uncertainty before the sale and directly support Customer Conversion
The Unboxing: What Four Customer Journeys Revealed About the CX Behind Creator-Driven Commerce
Journey Friction Matrix
Journey Stage
Legacy Collectible Jewelry
Accessible-Luxury Jewelry
Prestige-Portfolio Beauty
Mass-Retail Clean Beauty
Social Discovery + Engagement
Pre-purchase TikTok question went unanswered
Pre-purchase TikTok question went unanswered
Pre-purchase TikTok question went unanswered
Pre-purchase TikTok question went unanswered
Pre-Purchase Questions
Question ultimately answered through owned channels
AI provided substantive answers and explained membership pricing
AI provided a detailed product comparison with clear limits
Useful response, but email verification created access friction
Purchase + Fulfillment
Product arrived with damaged packaging and quality issues
Purchase led to confusion about an unexpected membership charge
Smooth purchase and fulfillment experience
Smooth retailer-managed purchase and fulfillment
Post-Purchase Support + Returns
Broken return portal and multiple follow-ups required
Straightforward return; membership charge ultimately refunded
Fast, straightforward self-service return
Straightforward retailer-managed return
Key:
Significant Friction
Minor Friction
Smooth Experience
Brand-by-Brand CX Breakdown
Each of the four brands demonstrated different strengths and gaps as customer intent moved from discovery to purchase, support, and potential repurchase.
The Legacy Collectible Jewelry Brand: Strong Marketing, Difficult Exceptions
This brand had a product proposition built around nostalgia, collaborations, and highly shareable products. But when the customer journey moved beyond the expected path, the experience became considerably more difficult.
What Worked:
- Ultimately issued a full refund for the damaged purchase
- Responded to our later concern through both Facebook and Instagram
- Acknowledged and escalated a sensitive concern rather than making an unsupported claim
What Didn’t:
- Initial TikTok pre-purchase question went unanswered
- Return portal failed when we attempted to use it
- Warranty workflow didn’t preserve our explicitly stated preference for a refund
- Refund resolution required multiple follow-ups
- Conflicting refund amounts created additional uncertainty
- Responses to the final purchase barrier acknowledged the concern without actually addressing it
Key Takeaway for CX Leaders:
Marketing can create product demand, but exception handling determines what happens when the customer journey doesn’t follow the happy path. Preserve context and resolution preferences across workflows so customers don’t have to repeatedly explain the issue or push it toward resolution themselves.
The Accessible-Luxury Jewelry Brand: Useful AI, Uneven Channel Coverage
This brand demonstrated how useful AI can be when it has access to the right information. Its virtual support experience answered questions that directly affected purchase confidence, including a sensitive concern about its paid membership.
The challenge was reaching that useful experience consistently.
What Worked:
- AI provided substantive answers to pre-purchase questions
- Virtual support clearly explained how guest and membership pricing worked
- AI directly addressed the concern preventing a purchase rather than avoiding it
- Post-purchase support identified and explained the unfamiliar membership charge
- Support ultimately canceled and refunded the membership charge
What Didn’t:
- Initial TikTok inquiry received no response
- Later Facebook and Instagram inquiries also went unanswered
- Membership structure created post-purchase confusion about an unexpected charge
Key Takeaway for CX Leaders:
Automation only creates value when customers reach it. Connect social, web, and support channels so the quality of the experience doesn’t depend on where a customer happens to ask the question.
The Prestige Beauty Brand: Strong Automation With Clear Limits
This brand delivered some of the most useful automated support interactions in the experiment.
Its virtual assistant handled both an ordinary product comparison and a much more difficult challenge involving contradictory information. Rather than pretending to know more than it did, it acknowledged its limitations.
The brand also chose not to accept Facebook messages rather than opening an inbox it wasn’t supporting.
What Worked:
- Detailed AI response to a competitive product comparison
- Strong self-service return experience
- Virtual assistant stayed engaged through difficult follow-up questions
- AI disclosed its virtual identity when asked
- Assistant distinguished between published information and what it could independently verify
- Facebook messaging wasn’t offered as an unsupported channel
What Didn’t:
- Initial TikTok inquiry received no response
- Later Instagram inquiry also went unanswered
- Contradictory website language created a trust issue the virtual assistant couldn’t fully resolve
- Pre-purchase automation was informative but sometimes formal and impersonal
Key Takeaway for CX Leaders:
AI can be useful without having every answer. Clear boundaries are more useful than unsupported certainty. But even strong automation depends on accurate, consistent information across the rest of the customer journey.
The Mass-Retail Clean Beauty Brand: Helpful Support Behind Unnecessary Friction
This brand produced one of the clearest contrasts in the mystery shop.
Once we successfully reached support, the responses were useful. The bigger problems came from the friction involved in reaching support.
What Worked:
- Substantive response to our pre-purchase product question once successfully submitted
- Bot correctly explained the retailer handoff when we asked about our return
- Facebook immediately routed customers toward supported service channels
- Instagram response to our sensitive-skin concern was specific and actionable
- Recommended relevant products and offered practical guidance rather than simply reassuring us
What Didn’t:
- Initial TikTok inquiry received no response
- First web support inquiry never reached support because email verification was required after form submission
- Verification requirement was easy to miss
- Automated support messaging included outdated self-service links that led to 404 pages
- Retailer handoff created additional complexity around who owned the post-purchase experience
Key Takeaway for CX Leaders:
Strong customer support loses value when customers face unnecessary barriers to accessing it. Audit the entire journey into support, including forms, verification steps, automated emails, links, and channel routing, not just the interactions that successfully become tickets.
CX Signals: Best-in-Class and Missed Opportunities
Support Element
Best-in-Class
Missed Opportunity
Social Channel Coverage
Mass-Retail Clean Beauty Brand
- Clearly routed customers to a supported channel when direct service wasn’t available
Legacy Collectible Jewelry Brand
Prestige-Portfolio Beauty Brand
Accessible-Luxury Jewelry Brand
- Purchase questions went unanswered on social channels that appeared open for customer conversations
Pre-Purchase Guidance
Prestige-Portfolio Beauty Brand
Accessible-Luxury Jewelry Brand
- Provided specific, useful product or purchase guidance that helped address questions before purchase
Mass-Retail Clean Beauty Brand
- Useful guidance was available, but unnecessary friction made support harder to reach
Automation Usefulness
Prestige-Portfolio Beauty Brand
Accessible-Luxury Jewelry Brand
- AI provided detailed, relevant answers to product and purchase questions
Mass-Retail Clean Beauty Brand
- Automated support messaging directed customers to outdated self-service links that returned 404 errors
Self-Service + Returns
Prestige-Portfolio Beauty Brand
Mass-Retail Clean Beauty Brand
- Returns were straightforward and easy to complete through self-service
Legacy Collectible Jewelry Brand
- A broken return portal forced the customer into a higher-effort support workflow
Exception Handling
Accessible-Luxury Jewelry Brand
- Support ultimately resolved the unexpected membership charge and issued a refund
Legacy Collectible Jewelry Brand
- Resolving a damaged-order/refund exception required multiple follow-ups before the full refund was confirmed
5 CX Moves Worth Taking From the Experiment
1. Connect Social Demand to Customer Support
If customers can message your brand on a social platform, decide who owns those messages and what happens next.
You don’t need live agents monitoring every channel around the clock. You do need a deliberate experience.
Monitor the channel, automate an appropriate response, redirect customers somewhere that is monitored, or disable messaging when possible.
Silence shouldn’t be the strategy.
2. Measure the Journey Into Support
Companies carefully measure what happens after a ticket is created. Pay equal attention to what customers experience before the ticket exists.
Test your own contact forms. Click every link. Complete every verification step. Try the return portal. Send messages from outside your corporate network. Follow automated instructions exactly as a customer would.
A customer who abandons the journey before creating a ticket won’t appear in your support metrics.
The absence of a ticket doesn’t mean there wasn’t a customer problem.
3. Give AI Operations an Owner
AI and automation need ongoing operational ownership after launch.
Someone needs responsibility for keeping knowledge current, testing workflows, reviewing failures, maintaining links, identifying recurring exceptions, and turning those exceptions into improvements.
Your business will change.
Your AI needs an operating model capable of changing with it.
4. Design for Exceptions, Not Just Happy Paths
Automate the common path wherever doing so improves the experience.
Then spend just as much time deciding what happens when the customer leaves it.
Define when humans should step in, what authority they have, what customer context follows the interaction, and who owns an issue that crosses systems or teams.
Customers shouldn’t become the project managers of their own support cases.
5. Train Support to Recognize Purchase Intent
A pre-purchase question isn’t simply another inquiry to close.
Sometimes it’s the final obstacle between interest and revenue.
Give agents and AI enough product knowledge and context to understand what the customer is really trying to decide. Answer the stated question, but also look for the concern underneath it.
The goal isn’t merely to respond.
It’s to help the customer move forward.
Turn Influence Into a Customer Journey That Converts
Influencer-led commerce can create demand remarkably quickly.
A single post can introduce thousands of potential customers to a product. Social platforms can collapse the distance between discovery and purchase from days to minutes.
But that also raises the stakes for the Customer Experience operation behind the brand.
When a customer stops scrolling and starts asking questions, someone needs to be ready to answer.
When automation encounters an exception, there needs to be somewhere for that exception to go.
When the business changes, the systems representing it need to change too.
And when a customer tells you exactly what’s preventing them from buying, that’s an opportunity to do more than close a support interaction.
It’s an opportunity to earn the sale.
SupportNinja helps fast-growing companies build Customer Experience operations that connect Customer Conversion, Customer Onboarding, customer support, Technical Customer Support, and Customer Renewals across the full customer lifecycle.
Because creating demand is only the beginning.
Your Customer Experience has to be ready to catch it.
