Tolstoy helped Primrose, an online retailer, tag and filter tens of thousands of customer emails.
Primrose is the UK’s largest online garden retailer. During the pandemic lockdown, they received an influx of orders from customers staying at home.
The company usually routes customer inquires through Freshdesk, an email platform, that helps categorize inbound mail. But they received a huge increase in direct email, which had no categorization applied. They were confronted with the prospect of reading through a backlog of 50,000+ emails — with thousands added per week — just to triage high priority emails, such as cancellation requests.
Tolstoy generated a custom model to tag the retailer’s emails, using their previously manually categorized messages as training data. We also helped them discover that customer-inputted categories via Freshdesk were often incorrect, with an accuracy of 50-70%. We corrected these for them.
Using Tolstoy’s model, Primrose was able to categorize their backlog of emails with 98% accuracy, 97-99% precision/recall. It took less than two days to set up the model, after which Tolstoy processed tens of thousands of emails in seconds. This would’ve taken an estimated 4-5 agents about a month to manually process.