How Lolahemp Cut Meta Cost Per Acquisition From $143 to $62.70
Lolahemp is a Shopify store selling vet-formulated CBD for dogs and cats, run by a small team. It had run Meta ads both in-house and through outside agencies before trying Agency AI. With Agency AI, its Meta cost per purchase fell from $143 to $62.70.
The Problem: A Paused Account and a $143 Cost Per Purchase
Between August 1 and September 30, 2025, a 61 day window, the campaigns in the account looked like this. By the end of that window, every one of them was paused.
- ASC - VSL: 4 purchases at $200.69 each
- ASC - Testing: 15 purchases at $161.55
- Shopping (2): 3 purchases at $120.08
- ASC - Prospecting: 11 purchases at $105.22
Together that comes to 33 purchases and a cost per purchase of $143.
Neither the team's own work nor its past agencies had brought cost per acquisition down to where it needed to be.
What Changed
Agency AI is a Shopify app that runs Meta and Google as one coordinated funnel. It builds a brand profile from the storefront, including tone, target customer, value propositions, and a break-even ROAS the merchant sets, then generates recommendations on a trailing seven day window. Each recommendation shows the proposed change, the numbers behind it, spend, ROAS, CPA, CTR and conversions, and plain-English reasoning measured against that break-even figure.
One point matters more than any other for a store owner reading this. On Meta, nothing is applied without approval. Agency AI does push optimization changes into a Meta account on the merchant's behalf. It pushes only the ones the merchant approves. Meta recommendation cards carry an Apply button, the same as Google's. When the AI Strategist chat proposes a Meta change, that proposal sits as pending until a person clicks Approve. Editing an ad ends at an explicit publish step. A campaign only goes live on Meta after a human at the store says so.
After the switch, two campaigns ran live, both carrying the "Agency AI //" naming prefix we use to bring campaigns into tracking: one top of funnel test and one retargeting test.
The Results: $143 to $62.70
| Before | After | |
|---|---|---|
| Window | Aug 1 to Sep 30, 2025 (61 days) | Oct 10 to Nov 6, 2025 (28 days) |
| Campaign status | All paused | Both active |
| Purchases | 33 | 95 |
| Cost per purchase | $143 | $62.70 |
Cost per purchase fell from $143 to $62.70, a 56 percent reduction.
Both new campaigns beat every campaign from before. Retargeting came in at $56.12 per purchase and the top of funnel test at $64.96, both well under the cheapest before-period campaign at $105.22. Volume moved in the same direction, with 95 recorded purchases in 28 days against 33 across the prior 61.
Both figures come from Meta Ads Manager, covering a 61 day window before and a 28 day window after. Our Vapor95 case study shows a similar result on Meta, with CPA down from $52 to $21.50, about 59 percent lower.
What the Lolahemp Team Said
"Agency AI has outperformed both our internal ad efforts and every agency we've ever worked with. Our cost per acquisition on meta is down over 25%, and the best part is how hands-off it's been. We'll be using the platform across all of our paid channels and recommending it to other store owners in our network."
The results in Ads Manager went further than the quote: cost per purchase fell 56 percent.
Why This Matters for Shopify Store Owners
Plenty of ecommerce brands conclude that Meta has simply stopped working for them. A common shape across stalled accounts is several overlapping campaigns, thin purchase counts spread across them, and a blended cost per acquisition that nobody has the hours to pull apart week after week. That is usually a structure and attention problem before it is a platform problem.
What a rules engine contributes is the attention. It reads the account on a rolling weekly basis, checks performance against the break-even ROAS you have defined rather than a generic benchmark, and hands you a decision with the evidence attached. You still make the call on Meta, every time. That is the design, and for merchants spending their own money it is the right way round.
If your Meta cost per acquisition keeps climbing and nobody has time to audit it, the Lolahemp numbers are worth a look.
Frequently Asked Questions
Sources
Performance figures come from Lolahemp's Meta Ads Manager account, for August 1 to September 30, 2025 and October 10 to November 6, 2025. The testimonial is from the Lolahemp team. Brand details come from lolahemp.com, and product behavior from our knowledge base and Shopify App Store listing.
Free to install. Launch campaigns and pay nothing for your first 30 days.
