Daki — Supermarket in minutes
A Product Discovery Case Study
What is Daki?
Daki was created with the idea of becoming a digital supermarket with up to 15-minute deliveries. The startup’s delivery approach is straightforward: small distribution facilities scattered across the city, serving a radius of up to 3 kilometers.
These mini-centers are known as dark shops, and they are scaled-back distribution hubs — exclusive locations for storing, sorting, and shipping items online.
In 2021, the business secured a USDS$260 million Series B investment. The firm was valued at USD$1.2 billion at the time, and was advertised as “one of the fastest startups to attain unicorn status in history.”
Current customer journey
1- Customers use the app to select items they wish to receive at home
2- An in-house crew at the dark stores assembles orders
3- Couriers on motor bikes pick up the selected order and the package is delivered in up to 15 minutes and free of charge
Despite the fact that Daki is presently expanding its commercial operation remains challenging. In particular, the checkout conversion rates. This article focuses on the process of identifying problems and exploring potential solutions. It begins by posing the question;
How can we boost the checkout conversion rate for Daki orders?
Preliminary hypothesis
This rate of shipping hypothesis was later explained by further investigating the business strategy and the choice to support the startup by redirecting funds from an investment round into a free delivery model. It is unclear whether Daki will continue to offer free delivery in the medium to long term. Given this knowledge, hypothesis 4 was declared invalid and was displayed in a different color on the opportunity map.)
The information is arranged systematically in the opportunity map displayed below.
Further understanding the scope of the problem
Given Daki is a B2C model, understanding customer preferences is vital to operations. Simply put, successfully serving the consumer drives customer loyalty and translates to revenue.
The methodological technique selected included a survey to test the validity of the preliminary hypothesis, followed by an interview process to further elaborate on details.
Following up on research the team would discuss results and outcomes, brainstorm over the gathered data and ideate with the goal of translating data into actions. The target is to get more customers using Daki services and reduce abandoned carts in the application.
At this point, a disclaimer is required: Given Daki is an existing organization, some information was unavailable. The lack of datasets required assumptions to be made. The target in this scenario was established arbitrarily at a 0.7% monthly increase in checkout conversions.
“Before I order, I’d want to know how many products are available. Another feature that I’ve only seen on Amazon is the ability to schedule recurring purchases of things that I buy regularly; grocery apps could offer this as well.”
Research process
Conducted with the objective of learning about respondents’ preferences and determining where adjustments are needed. This initiative discovered the following.
“I don’t always find what I’m looking for.” Express delivery in my city does not contain the same goods as scheduled delivery.”
The survey
In designing the questionnaire, the assumption was that a broad inquiry could reveal concepts that had been overlooked. Due to the search for broad information and personal preferences, no specific screening criteria were used to determine participation.
Daki only operates in four Brazilian metropolitan cities, which may have resulted in biased information. This premise was addressed by making the questions screenable and filterable during analysis.
What product categories do you typically purchase with express delivery apps (up to 15 minutes)?
The top 5 categories — Beverages (alcoholic and non-alcohlic), Pharmacy, Hortifruti (fresh fruits and vegetables), Mercearia (non perisheables) and snacks.
If this were not a case study but a real life assessment, these self-reported categories would have been triangulated to Analytics data (GA), considering this is a case study and limited information was available, the self-reported categorization was taken as reliable evidence.
Findings on overall user experience
Concerning the key problems, respondents claimed that there are issues in three aspects: assortment, filters, and usage of discounts coupons.
Self reported usage pattern
- Respondents report that they typically buy 5 to 10 items.
- For ‘two persons, including me’.
- Shopping frequency: Every month.
- They understand that the shopping experience is good.
- Receiving a discount coupon through push notification is a motivator for 64% of respondents to complete the purchase.
Turning data into actionable insights
After outlining challenges, conducting more research on client requirements, and brainstorming on desirable outcomes. Three key strategies and actions were outlined, with the goal of increasing checkout rates.
- Increase the number of products and brands available.
- Discount coupon for recurring purchases or for a favorite product.
- Improve the filtering and grouping design to increase visibility.
Roadmap
These steps should be partially or fully implemented within three months.
Assortment
To expand assortment of brands and products the suggestion is to expand collaboration with supermarkets in close range to the dark stores. This iniciative would be complimentary to the current business model and tested with selected products and implemented associated to promotional campaigns. Larger supermarkets chains may have more complex logistics and a longer time horizon for shipping merchandise.
Geolocalization intelligence and user targeting would be factored in to determine the adequate campaign, collecting MAU (monthly active users) metrics prior to the campaign and after.
Layout
In the layout improvement plan, users have complained about poor classification and suggested the possibility of a swipe up menu to display a description of pictures, nutritional features, and characteristics of the goods.
Layout enhancements could be A/B tested to assess whether whey are adequate to engage customers. This could also minimize efforts in development. Moreover, the data can provide sufficient evidence to support coding efforts.
Filters and coupon usage
The third change is to add a feature that allows customers to store their favorite goods so that they can easily discover them later on without having to scroll through the menu of options.
A few extras
- To minimize churn, when it comes to discounts and uncovered areas, present the client with an objective and clear explanation or redirect flow automatically.
- Push communication to alert when the desired item is restocked
- Design a favorite product feature for future purchasing.
- Determine the customer satisfaction index via Likert scale or similar assessment feature
Metrics employed in the validation process
- Assess daily active users (DAU), weekly active users (WAU), and monthly active users (MAU) throughout the roadmap and primarily at the end of the three-month process. Compare values to their present state to see if there was an increase.
- At the end of the process, suggest a Likert scale evaluation or other type of fast survey to consumers who were engaged during the implementation phase in exchange for a discount coupon.
- Determine whether product promotions raised client interest in purchasing presently available products. Analyze the purchase volume of checked-out carts that include promotional items.
Next steps
If the initiative is successful, there should be a whole new set of emerging challenges that can provide the opportunity to further iterate and offer an improved service to the final customer, based on the perspective of Continuous Discovery.
This article is a case study for Tera’s Program in Product Discovery in September 2022. Although Daki is an organization, the author is not affiliated with it and does not work for it. The suggestions for improvement pertain to a case study and have not been presented to Daki’s team.
