How Puzzle Financial increased their subscriptions by 3.9x & reduced their CACs by 58% in 90 Days

Puzzle Financial is an AI accounting platform built for forward-thinking startup founders, CFOs, and finance teams. Puzzle Financial sits in a crowded category. Their buyers already know tools like QuickBooks and legacy accounting software. Puzzle’s edge is speed and automation – think “AI accounting that turns hours of reconciliation into minutes.”

Because of that competitive context, Puzzle relied on:

• Partnerships and brand
• Google search demand
• Paid acquisition to drive high-intent demos and self-serve signups

When we first spoke, their mandate was simple:

“Growth was flatlined. We were driving volume but not quality: onboarding and subscriptions stayed flat, CPCs were high, budgets were inconsistent, and our previous consultant couldn’t explain sudden metric drops. I needed a fresh approach to reset Google, activate Meta and LinkedIn, and build a predictable, efficient acquisition engine that could actually spend and grow.”

Within 90 days of working together, we:

• Cut paid media CAC by roughly 90% vs their pre-2x benchmark, and over 60% vs the most recent baseline
• Lowered average Google Ads CPC by 68%
• Grew paid subscriptions by 19.9x year-over-year
• Increased signups by 2.6x across Google + Meta
• Stabilized monthly ad spend so the team had a predictable pipeline

When Puzzle came to 2x, they were stuck:

• ‍Flat account creation volume was flat — growth had stalled heading into their peak acquisition season.
• Customer Acquisition Costs (CAC) was painfully high — they needed to double customer acquisition but couldn't hit efficiency targets.
• Paid social (Meta) was underutilitized  — due to a lack of confidence in the channel, creating a critical gap in top-of-funnel acquisition.
• Monthly spend was wildly inconsistent — monthly spend fluctuated wildly, making predictable scaling difficult and leaving allocated budgets underutilized.

We started with a deep audit of Google Ads, Meta, and the conversion flow from ad click → signup → onboarding → subscription.

Here’s what surfaced.

• Broken conversion tracking — platform events didn't match their primary source of truth, causing inaccurate reporting and mistrust of in-platform metrics.
• Bidding strategies were set to overspend — poor tracking led to overbidding and inflated CACs.
• Google Ads underperforming — campaigns had inefficient structures, outdated ad copy, and settings that limited reach and targeting.
• Landing pages had QA issues — broken links and UI frictions on mobile which represented big part of their traffic
• Sign-up flows had UX friction — confusing account creation, dead ends, and missed conversions

Individually, none of these problems seemed immediately clear of their negative impacts but together, they explained why Puzzle was paying so much for each customer and getting so little back.

Instead of chasing one “hero campaign,” we focused on building a repeatable system Puzzle could rely on.

Step 1. Fix the measurement foundation

First, we aligned on what actually matters for growth:

• Signups – people raising their hand
• Onboarding completes – users who set things up seriously
• Subscriptions – paying, retained customers

We then:

• Consolidated campaigns down to one primary optimization event (Onboarding Complete) so the algorithms could learn from a clean signal that was both close to revenue and generated enough volume for stable learning.
• Cleaned up conversion tracking so that internal dashboards and platform reporting finally told the same story.

Result: when we made changes in Google and Meta, we could trust the impact we were seeing.

Step 2. Reactivate Meta as a proper demand engine

Next, we turned Meta from “unused line item” into a working growth channel.

a) Start with customer insight, not ad formats

We interviewed the sales team and dug through their best current customers:

• Why did they switch to Puzzle?
• What pain felt urgent enough to move away from spreadsheets and legacy tools?
• Which features actually show up in sales conversations?

Then we built creative around those real problems and benefits:

• “Still using spreadsheets?” style FOMO angles
• “From manual to modern” positioning
• Messaging tailored to founders, CFOs, and finance teams separately

b) Let Meta find the right buyers

Instead of obsessing over hyper-narrow targeting, we:

• Fed Meta high-quality creative tied to clear pain points and outcomes
• Let the algorithm find similar profiles to their best customers
• Used multi-creative test cycles to quickly identify which angles pulled in signups and downstream subscriptions

c) Build remarketing that adds proof, not just frequency

Once the top-of-funnel was humming, we layered in:

• Credibility ads (social proof, “who trusts Puzzle,” awards, etc.)
• Product explainers to reduce uncertainty for evaluators
• Simple retargeting that nudged warm visitors back to sign up or book a demo

Meta no longer lived in a silo. People who discovered Puzzle via social would later search for the brand on Google, reinforcing performance there too.

Step 3. Rebuild Google Ads around intent and efficiency

With measurement fixed and Meta activated, we tackled Google.

a) Separate branded and non-branded campaigns

We split Google Ads into:

• Branded campaigns – users already searching “Puzzle” (warm, high intent)
• Non-branded campaigns – people searching “AI accounting,” “AI bookkeeping,” “accounting software for startups,” etc. (cold-to-warm intent)

Each bucket got its own structure, bidding strategy, and budgets. But the key to make sure this worked was having a process for ensuring negative keywords and brand exclusions were put in place.

b) Dial back over-aggressive bidding

• For branded, we moved from an aggressive target CPA strategy to more controlled manual CPC bidding.
  • We lowered bids to what was actually required to win the auction on their own name, not what the algorithm assumed it could extract.
• For non-branded, we recalibrated target CPA and rebuilt ad groups around tighter, more relevant keyword themes, backed by the same customer insight we used for Meta.

This combination:

• Dropped CPCs by 68% within two weeks
• Freed up budget to reach more net-new people at the same (or lower) spend

c) Tighten the click-to-signup experience

Alongside campaign changes, we:

• Partnered with the Puzzle team to align landing page headlines and copy with specific keyword themes (e.g. “AI accounting for startup finance teams”) with each core concept having their own dedicated landing page.
• We worked with the team to clean up any UX friction through QA passes on mobile and desktop

This ensured visitors from each ad group landed on a page that matched their intent, not a generic homepage, with a sign up process that work as intended.

Once the system was in place, the numbers moved quickly.

1. CAC down, subscriptions way up within 90 days

• 58% reduction in paid CAC vs their pre-2x benchmark
• 3.9x growth in subscriptions during the measured window

2. CPC and efficiency improvements in Google

• By optimizing campaign structure and bidding strategies, we reduced Google Ads CPCs by 68% within two weeks — a key driver in cutting overall CACs.
• This resulted in more clicks from the same spend, and more of those clicks turned into meaningful actions (signups, onboarding completes, subscriptions)

3. Signups and top-of-funnel growth

Comparing the May–July baseline to August–October (after 2x took control):

• Signups increased by 2.6x across Google + Meta
• Google signups grew by 179%
• Meta signups grew by 1,900% as we turned a dormant channel into a consistent demand source

4. Stable spend, predictable pipeline

• By improving their Google Ads campaigns bidding signals, and activating their paid social ads on Meta, we were able to stabilized their spend month over month enabling predictable scaling and consistent lead volume.
• Both channels worked together: Meta created demand, Google captured it, and internal reporting finally matched what the platforms were saying

Why this matters if you’re a B2B SaaS founder or marketing lead

There wasn’t a single “secret” tactic in this project.

What changed the outcome for Puzzle was:

• Clean measurement – so you can actually trust CAC and have the team and leadership fully aligned
• Channel activation with a clear process – Meta wasn’t scary once we had tracking, creative strategy, and system in place
• Intent-based Google structure – branded and non-branded served different jobs and their performances were no longer mixed
• Consistent budget and strategy – less guessing, more learning and compounding

If your situation sounds similar — flat growth, expensive customers, inconsistent paid performance, and a social channel you “keep meaning to figure out” — the Puzzle playbook is repeatable.

What we did for Puzzle, we do for other B2B SaaS brands

We helped Puzzle:

• Rebuild their measurement so they could trust the numbers
• Launch and scale Meta as a true demand engine
• Restructure Google Ads around real buyer intent
• Cut CAC dramatically while growing subscriptions and stabilizing spend

If you’d like to see what this looks like for your own acquisition engine — book a call today.

We’ll review your current Google & Meta setup, walk you through where money is leaking today, and share what we’d prioritize in the first 90 days to move you from “expensive and unpredictable” to “scalable and measurable.”