Skip to content
Agents tracked: 492 Releases (7d): 545 Agent status: 1 with issuesSince last limit reset: Claude 9d · Codex 7d For AI agents: Markdown Updated Nov 20, 2026
Case study Bared FootwearResults as reported by Sierra

Bared Footwear resolved 30% of email and 52% of chat with a Sierra agent

Shoe brand Bared replaced a decision-tree chatbot with a Sierra agent that takes actions, not just answers: resending gift cards, issuing return labels, checking orders and applying sale credits. Its own support team builds the agent's journeys. Sierra reports the agent resolves about 30% of email and 52% of chat two months after launch.

The problem

Thousands of emails arrived every week and every one was answered by hand, many of them simple requests such as resending a gift card, sending a return label or checking an order. Around its two flash sales a year the inbox backed up, and extra staff could not be added quickly because product knowledge takes months to learn. Live chat ran on a decision-tree bot where customers got stuck and a single wording change meant editing several branches.

How they did it

  1. Set a clear requirement before choosing a tool, in Bared's words an agent that can "take an input, chain together a series of actions, and return a response to a customer", not one that only writes replies.
  2. Launch on chat first, plus a few email journeys, rather than every channel at once (Bared went from kickoff to live in five weeks).
  3. Let the support team own the journeys. Bared's customer service automations lead and the team build and maintain them and add at least one new journey every week (FAQs, shoe care, sizing and fit, returns and exchanges).
  4. Automate a policy decision end to end. In Bared's sale journey the agent checks the purchase date and whether the item is still in stock, then issues store credit for the price difference with no team member involved.
  5. Decide in advance which cases must go to a person, and report those planned handoffs separately from cases where the agent could not find an answer.
  6. Have the agent gather the order number and context before handing a case to a human, so the team can act on it straight away.

Results as reported by Sierra

  • The agent resolves about 30% of email volume and 52% of chat two months after launch, and both figures climb weekly, as reported by Sierra.
  • Kickoff to live took five weeks for chat and some email journeys, as reported by Sierra.
  • The 300-email backlog that built up around sales is gone, and complex cases reach the team pre-qualified, as reported by Sierra.

These figures are as reported by Sierra in Sierra customer story ↗ (Oct 22, 2026). Sierra sells the product used here. AgentGid did not measure or independently verify them.

Takeaway. An agent earns its keep on support when it can complete the simple transactions itself and when the team that knows the customers writes and owns its journeys.
AgentGid's take

This suits mid-size retail and e-commerce brands with heavy, repetitive email (order status, returns, gift cards) and a support team willing to maintain the agent's journeys every week. It is not a fit for small shops that want a self-serve tool with a listed price: per AgentGid's Sierra page (pricing checked Oct 5, 2026), Sierra has no public price list and no free tier, pricing is outcome-based and must be requested from sales, and its documentation sits behind a login, so you cannot try it before talking to the vendor. The figures are Sierra's own, from a customer story on its site, and are not independently verified; the 30% email figure is also early, two months after launch. If you need something you can run and inspect yourself, Rasa and Chatwoot's Captain agent are listed as alternatives in our catalog, though both need more of your own build work for actions like issuing store credit.

The agent used here

Similar use cases

Guides