Update - Week 2
14 days since the beginning of the journey.
Progress has been made and we’re slowly, but surely approaching our first milestone.
Despite the promising results, we did encounter a few problems during the weekend that pushed us a step back. Fortunately, we managed to resolve them within 24 hours and we’re back on track!
Let’s first look at the change we achieved in terms of slaves, follow actions and DMs sent.
| Week 1 | Week 2 | Change |
Active Slaves | 433 | 477 |  10%
|
Avg Daily Follow Actions | 33.365 | 52.403 | 57%
|
Avg DMs Sent | 2.187 | 4.127 | 88%
|
And here’s the daily performance:
Daily Follow Actions
Week 1
Week 2
Daily DMs
Week 1
Week 2
As you can see there’s a visible shift during the weekend of Week 2 in both follow actions (drastic decrease) and sent DMs (drastic increase).
The discrepancy in Follow Actions and DMs is due to two unrelated factors.
Let me first address the drop in Follow Actions. At the beginning of Saturday, we refreshed and cleaned our source list. We currently have over 1.500 sources (adding more daily), most of which are accounts between 10.000 and 50.000 followers. When we refreshed the list, the automation software started looking for accounts to follow sorting them from the most recent followers. Since the majority of those followers were already followed, it needed to “dig” much deeper to reach the users that haven’t been followed. Additionally, there was downtime on half of the proxies, which certainly didn't help. Based on the observations, this issue took around a day and a half to sort itself out. Starting today, we’re back where we stopped on Friday.
As for the drastic increase in sent DMs, it was due to a test where we changed who we sent DMs to.
Until now, we were only sending DMs to people who we previously followed. That meant that whoever followed us "organically" wouldn’t receive a DM. Since some of our slave accounts are already thousands of followers big, we decided to test sending DMs to “organic” followers as well, to see what kind of results that will bring. As predicted, this resulted in a massively increased number of DMs, however that didn’t bring any improvement in our in-store performance. Quite the contrary, combined with the decrease in follow actions, most of the DMs sent was to these followers who didn’t come from our provided sources. According to what we saw, the vast majority of these DMs were sent to bots who followed our accounts, so we decided to stop this. We are now back to the previous approach, and the number of DMs should be back to the regular threshold from the previous work week.
Onto the store metrics:
| Week 1 | Week 2 | Change |
Sessions | 299 | 475 | 58% |
Sales | 2 | 8 | 300% |
ATC | 14 | 60 | 328% |
Conversion rate | 0.67% | 1.68% | 150% |
AOV | $36.27 | $27.73 |  -24% |
Revenue | $72.54 | $221.82 | 205% |
As you can see the tests that we did at the beginning of last week were fruitful. The split-testing showed clear winners that substantially increased the number of sales, ATCs, conversion rate, and revenue. The only metric that dropped was the average order value, but that was expected, as our estimated AOV is around $28. We’ll work on improving this in the future by introducing cross and upsells, but that’s not a priority right now.
We also noticed a clear pattern when it comes to the conversation with potential customers. The days that we got the most sales were days in which we were most responsive to customers, i.e. we answered new messages within minutes of the customer writing. In contrast to that, on the two days in which we were absent for the majority of the day and answered after 6+ hours, we saw no conversions at all, despite the number of messages we received being the same. Of course, this could include a multitude of factors, but we saw the same thing repeat twice, so we’ll pay a close eye to this and try to be as responsive as possible.
As soon as we are in the position to pay someone, we’ll hire an assistant to work 6-8 hours per day and mostly respond to messages from potential customers.
This insight into the behavior of customers also inspired me to rethink my PPC approach, and I’ll be testing some campaigns for e-commerce clients in the future not with the conversion goal, but rather with the message goal and then have someone answer people on Messenger/Instagram and try to convert them into customers. From my current analysis, message campaigns will get you at least 50% higher reach than conversion campaigns with the same budget, and with the right funnel, this could be a goldmine in itself.
Finally, we’re currently working on increasing the quality and trustworthiness of the brand. We are negotiating a batch of 500 or 1000 branded package boxes that we will ship the units in. We are also doing outreach to micro-influencers that are willing to create images and videos for us that we can use for our Instagram page and website, so it looks cleaner and more legit.
That’s it for now and I’ll see you again next week!
) made a thread with same method & goal back in 2018! for sure its been hard since last API IG (july2019) decrease update, but sure still working! good luck!
