2 Ways To Make Your Business More Profitable During & After COVID-19
Your medium-to-large size business still has a chance to survive the recession…
Feels like the world is on fire, doesn’t it? People aren’t being allowed outside, local businesses are shutting down, and you’re wondering what’s going to happen to your company.
Your first strategy was to lean down your company: cut costs. It’s a rational decision, considering that you’re uncertain of this year's revenue stream(s).
Nonetheless, you know that cutting costs aren’t enough.
You need to figure out how to optimize your business for profitability and double down on those revenue streams. There’s no longer any more time for “guesses” and “gut-feelings”, you need to hardcore facts and data to base your decisions on during and after COVID-19.
In this article, I’m going to give you some example scenarios of how a business can increase profitability during this time. Though these examples are based on previous results we’ve provided, I choose to use a fictitious company name (i.e. ExampleRetail.com) in order to respect the privacy of our clients at Silicon Valley Consulting. We normally primarily work with AdTech companies, this was our initial attempt branching out.
Optimize Paid Ads
The online luxury costume store, ExampleRetail.com, is looking for how to increase their profit margin. They spend $2M per month on Facebook ads. Thanks to this ad spend, they drive around 1M potential customers to their site (roughly $2.00 per click).
About 1% of those monthly 1M Facebook ad clicks actually turned into paying customers: ~10,000 sales per month. The average order amount is $350, which equals $3.5M dollars per month. After advertising costs, but before manufacturing/misc. costs, they’re making $1.5M Return on Investment (ROI) and a 1.4 Return on Ad Spend (ROAS) a month.
They decided to bring us on as an external consulting team of Machine Learning Engineers and Data Engineers. During the initial pilot stage (1–2 months), the team was told to explore ExampleRetail.com’s advertising data. The data was spread out across many sources: from Shopify, Facebook, Google Analytics, Google Sheets, etc. Luckily the team is used to this!
These were the insights that the team provided (*slightly tweaked for privacy):
They were mainly doing two approaches to paid advertising on Facebook:
- 1% Look Alike Audience (LAA) on all of their customers that purchased at least three times over the past 12 months.
- Targeting fans of Gen Z celebrities with over 200M followers
Our team suggested that ExampleRetail.com should:
- Try out their 3–4% LAA and 6–9% LAA on all of their customers that purchased at least three times over the past 6 months.
- For their 1% LAA, create new advertising creatives/content
The 1% LAA (basically, people who are nearly exactly to their existing customers) was being exhausted. At the time of analysis, each person in the audience segment had seen their existing adverting content roughly 10 times.
By looking through the historical data, the team was able to confirm that ExampleRetail.com’s click-through-rate (CTR) for an ad would start decreasing after every Facebook user in their audience saw that ad at least 7 times.
This doesn’t mean that an ExampleRetail.com ad wasn’t profitable after the ad was seen at least 7 times, it simply means that the cost-per-click (CPC) would increase.
Our team determined that if they completely swapped out their advertising content every 3 to 4 days, then they would potentially increase their current revenue and profitability by 10%.
This could be done if they hired another employee to handle advertising content freshness, but we decided it would be more cost-effective for them if we built out a system for them: a system that would automatically swap out ad content before CTR dropped below 2% or ROI dropped below $400.00 per purchase.
Additionally, we determined that their 3–4% LAA and 6–9% LAA are untapped audiences with similar ROI potential to their 1% LAA.
- Target Gen Z celebrities who specialize in cosplay
After analyzing the customers who purchased from ExampleRetail.com, specifically the one who bought after seeing an ad targeting fans of Gen Z celebrities, we determined that the majority of these customers are interested in cosplay.
If ExampleRetail.com targeted the cosplay interest on Facebook, it would be able to expand its current addressable market by 43%. This is based on the data trend that showed that people interested in cosplay are highly likely to buy from them, even the ones who aren’t interested in famous Gen Z celebrities.
We then offered to build a machine learning-powered dashboard that would allow them to automatically get the top-related Facebook interests for any sub-segment of their customers, then automatically A/B test which of these interests performed the best.
Optimize Customer Churn
ExampleRetail.com decided that as our Data Engineer built out their ad content freshness system, they would like to run a second pilot with us around their customer data.
Over the past 12 months, they were having a hard time increasing their customer value and life-time-value (LTV). On average, each of their customers has a customer value of $650 per year. However, the average ExampleRetail.com’s customer has a short customer life span of 2 years (short for a luxury retail brand), so their customer LTV is $1,300.
Going into this pilot, we decided to focus on two search objectives:
- Find data trends that would increase their average customer value
- Find data trends that explain why their customer lifespan was so short
After 1 to 2 months of data exploration, we found several methods to increase their customer LTV
- Improve customer personalization
Email campaigns were driving traffic regularly to their website: producing a positive ROI. However, we noticed a pattern where certain subsegments of their customers would engage more frequently with certain emails than others.
After some analysis, we came to the conclusion that ExampleRetail.com was not personalizing their email campaigns enough. We realized that based on the images in the email and the topic they chose for the email, it would engage certain demographics 2x more than average (e.g. choosing images that look like that audience subsegment). This would guarantee an increase in customer value and potentially increase customer lifespan.
We then offered to build a machine learning-powered dashboard for their Head of Email Marketing that would automatically build a profile of each customer based on various data sources, group similar customers, and send out personalized emails for each group.
Additionally, we found a similar problem on their website, where certain demographics would purchase more frequently based on the model in the product image. Based on what we saw with email marketing, we decided it would be best to build a tool that would allow their website to automatically change product imagery based on the internal profile that we built.
- Integrate SMS marketing and Micro-Influencer Marketing
We discovered that a high percentage of ExampleRetail.com’s customers were engaging with a niche group micro-influencers in the cosplay industry. Through manual inspection, we saw that this niche group of micro-influencers choose to engage their fans (i.e. ExampleRetail.com’s past, current, and potential customers) through SMS marketing.
We determined that if ExampleRetail.com engaged these micro-influencers (whether it be a paid sponsorship or a free creative collaboration), they would be able to get back the customers they’ve lost, be top of mind for existing/active customers, and get new customers.
However, (re-)obtaining this demographic wouldn’t be enough. Based on the trend we saw, it would be highly advised to do some test campaigns with SMS marketing since the aforementioned micro-influencers already proved that SMS marketing is the preferred method for their audience. This would increase the customer lifespan for ExampleRetail.com.
We agreed that once they run these sample SMS campaigns, we can do a follow-up pilot in order to focus on increasing the ROI of SMS campaigns through methods such as Machine Learning-powered personalization.
Conclusion
We can conclude that the two pilots were a success. We are currently in the second part of our contract with them and a few other clients: consisting of building machine learning-powered dashboards and data tools that help them capitalize on the insights we provided.
Even if you’re not an online retail brand if you’re interested in learning more about how the Silicon Valley Consulting team can help, check out: www.SiliconValleyConsulting.io
