01Recommend
Your next move, ranked by confidence.
Top recommendation · #1/5
Increase Shopping · All Products budget by 20%.
Strong evidence 95% CI
+$1,180 / mo
Know which campaigns are actually profitable.
Our Catalyst Audit grounds every recommendation in your first-party order data, ranks it by statistical confidence, and compounds with each audit you run.
02Validate
Track where your revenue truly came from.
Attribution match 88.7%
#1027 · $84.00VIA GOOGLE · SHOPPING matched
#1031 · $156.00VIA GOOGLE · SEARCH matched
887 of 1,000 orders match
03Measure
The margin that ad platforms can't see.
Profit ROAS · after COGS
2.21× median
92% probability above break-even
Trusted by brands investing in smarter performance marketing
Horizontally scrollable list. Swipe to see additional logos.
Search engine marketing gets more intuitive when you can credibly estimate the impact ahead, see what's actually influencing every campaign, and know the why behind every change.
Profitability · F-2
You think your ROAS is healthy. Once you layer COGS, a different story emerges.
Trellis applies your COGS to each attributed order and ranks every campaign by what it actually contributes from a margin perspective, not by what it looks like in isolation. Some "top-performing" campaigns may land below break-even once your margin lands. You'll know exactly which ones.
After COGS · ranked by margin
Brand | TM Search 6.2× ROAS +$4,680
Performance Max | Catalog PMax 3.6× ROAS +$3,744
Shopping | All Shopping 2.6× ROAS +$2,301
Display | Retargeting Display 3.1× ROAS +$1,248
NonBrand | Generic Search 1.6× ROAS −$663 positive ROAS, negative margin
Attribution · F-3
Platforms claim credit for orders. Your first-party data surfaces what actually happened.
Ad platforms may conflate clicks, views, and page visits as a primary conversion metric. Your store records what was actually purchased. Trellis reconciles every platform claim against your first-party order data and surfaces the duplicates, view-throughs, and visits platforms wrongly counted as sales. You'll know which sales were real.
Conversion ledger · 7d window
Order
Channel
Revenue
Source claim
Reconciled
#4218
Shopping
$89
Google
Verified
#4287
Search
$112
Microsoft
Verified
#4302
Performance Max
$189
Google
Verified
#4244
Free listing
$67
Google
Reclassified · free, not paid
#4225
Retargeting
$156
Meta
Duplicate counted twice by Meta
#4318
Search · Brand
$76
Microsoft
Verified
#4196
Search · Generic
$94
Microsoft
Reclassified · refunded net revenue $0
Across Google Ads, Microsoft Ads, Meta Ads → Canonical record: Shopify
Changelog · F-5
Your Changelog gives the next audit evidence to recommend what's working in your accounts.
Keep a record of the changes you make (budgets, campaigns, ad groups, and keywords) using our Changelog templates. During your next audit, the Changelog will evaluate which adjustments improved performance and which ones resulted in wasted spend. By your sixth audit, Trellis will be optimized and fine-tuned to your business, with recommendations that continue to build on each other over time.
Account changelog · 90d window
Mar 19
Bid strategy changed Shopping | Best Sellers
+12% ROAS Confirmed
Feb 28
Audience signal added PMax | Catalog
no lift detected Held
kept for next audit · evidence below threshold
Operator-recorded · 184 entries / 90 days → Read by next audit
Three steps from connection to compounding evidence.
01
Connect your accounts in minutes
Plug in Google Ads, Microsoft Ads, Meta Ads, and Shopify. Trellis reads performance and order data in read-only mode. You stay in control of every change.
Connects to
Google Ads
Microsoft Ads
Meta Ads
Shopify
02
Run a structured audit against your real economics
Trellis applies your gross margin, cross-references platform conversions against your store's orders, and gates every recommendation against a confidence threshold. The case for and against every major change is part of the report, not an afterthought.
Your business economics
AOV $58.24 CLTV $459.17 MARGIN 62%
03
Build a decision trail that compounds
Each cycle adds context to the last. At month one it's a log. At month six it's institutional memory: a record of what changed, what worked, and what to try next.
Apr 18 Retargeting
Launched 30-day cart abandoner audience
+$2,140 / 14d
Know what's earning, what's losing, and what to fix first.
The Trellis Health Score reads each connected ad account and rolls them into one profit-weighted number. Not a vanity metric. It surfaces your contribution margin (revenue derived from campaigns less ad spend and COGS), shows you the weakest pillar from your account, and recommends what changes to make.
Account Health · last 30 days
Health Score
89
/ 100
Composite of 5 pillars
Attributed Revenue
$20,397
+3.7% vs prior 30d
Contribution Margin
$6,884
−2.9% vs prior 30d
Blended ROAS
3.3×
+0.2% vs prior 30d
Spend
$6,374
+7.1% vs prior 30d
4 audits covering 30 of 30 days
Each recommendation passes five independent gates before it surfaces. Miss any threshold and the audit holds the finding for monitoring. You see what's worth watching, not what's forced from thin data.
- Statistical weight: enough conversions that the pattern isn't random.
- Audit depth: enough history to compare against your own baseline, not someone else's.
- Business context: your gross margin, your AOV, and attribution-validated orders.
- Falsifiable structure: each recommendation names a credible alternative and the threshold that would shift the call.
- Evidence traceability: every figure traces back to a value computed from your data, not a number a model invented.
When the evidence is weak, the audit holds the finding for monitoring instead of pushing a recommendation. A held finding is a valid result, not a failure. When the evidence is strong, the recommendation shows its work: the recommended direction, a credible alternative, and the threshold that would shift the call in the next audit.
The Changelog stitches each audit to the next, anchored by every change you record on the ad platforms. After three audits, the model projects budget impact. After six, it runs full inference calibrated to your account, not to a generic benchmark. The hours you used to spend combing reports and joining CSVs stay reclaimed.
A full audit written for your account, not a generic template.
Every audit applies your gross margin, cross-references your order data, and gates every recommendation behind a statistically significant conversion floor. The methodology is the rigor.
Whether you're an owner-operator, an in-house marketer, or an agency with 15 clients, Trellis calibrates depth, focus, and recommendations to your business economics. One methodology, adapted to how you work.
Owner-operator
1 of 4 unprofitable
−$340
Margin after COGS · 30d High confidence · 92%
+$3,420
Revenue$4,620
After COGS$3,420
Loss−$340
Clear answers, no spreadsheet required.
Know which campaigns earn after COGS and which lose money. Less time in dashboards, more confidence in every decision.
In-house marketer
Read into Audit #5
Changelog · Mar 26 R-228
Budget raised
Brand | Search · $80 → $96
30d outcome +$540 margin
Track every change to the revenue it produced.
The Changelog stitches your decisions to the outcomes they earned. Six audits in, your history compounds into every next recommendation.
Agency
Consistency 100%
Same methodology 4 client snapshots
Acme Co
3.2× ROAS
verified · COGS
Bayard
4.1× ROAS
verified · COGS
Coral Lab
2.8× ROAS
verified · COGS
Delta Goods
3.9× ROAS
verified · COGS
Same audit discipline across every client.
One methodology, identically applied. Consistent answers, consistent rigor, no per-client improvisation as your roster grows.
Make your next budget decision with answers, not theories.
Every audit provides a forward-looking report grounded in your margins and order data. Numbers you can stand behind and recommendations you can defend.
Annual billing saves 15%
Core
Four audits a month, so the next decision is grounded in last month's evidence.
Recurring audits grounded in your actual gross margin.
$169 / month
billed annually
Core includes:
- Every connected ad account
- Four audits each month
- Shopify integration with per-variant COGS
- Changelog of every change to your account
- Confidence-gated recommendations
- 90 days of audit history
- Datamart trends as a $99 / month add-on
Pro
Eight audits a month, twelve months of trend history, and a record of which changes paid off.
Track what changed, and what happened next.
$419 / month
billed annually
Everything in Core, plus:
- Eight audits each month
- Datamart with twelve months of trend history
- Recommendation outcome tracking — see whether each change worked
- Scheduled audits on a cadence you choose
- Custom COGS upload to override product-level costs
- 12 months of audit history
Agency
Audits scoped per client, with one roster for the whole book of business.
Consistent methodology across every client.
$339 / account / month
3-account minimum · billed annually
Everything in Pro, per client — plus:
- Eight audits per client each month
- Datamart with 24 months of trend history per client
- Per-client data isolation — separate setup, audits, and credits drawn from your agency pool
Five moments from a Catalyst Audit.
An audit starts with what the platform reports, traces the decisions that got you here, looks at what your store actually recorded, and ends with what to do next. The number in the middle is the recognition moment — the scope of what most teams act on without checking. Each card below turns a dashboard number into a decision you can defend.
COGS-Adjusted Profitability
Shopping — Promotional
Platform ROAS
3.1×
After COGS
1.1×
Break-even
1.0×
[FACT] SKU-level margin analysis
The margin story the platform doesn't tell.
A Shopping campaign reporting 3.1× ROAS appeared to be a top performer. The audit overlaid SKU-level margins and revealed the campaign was heavily weighted toward promotional products — contribution margin was well below the account average. What looked profitable at the platform level was operating near break-even after actual COGS.
85%
of ad spend decisions are based on platform-reported data alone — without first-party validation.
Platform data
ROAS3.1×Conv.197CPA$18.40
Changelog-Driven Detection
Change log Search — NonBrand
Bid strategy changed
Manual CPC → Target ROAS @ 4.0×
Device modifier adjusted
Desktop +15%
CPC +98% Conv −31% Spend +$840
Every change tracked to its measured outcome.
CPC doubled over six days with no explanation in the platform UI. The changelog traced the shift to a bid strategy change followed by a device modifier adjustment two days later — cause and effect, with timestamps.
Attribution Verification
Attribution check 30-day window
197
Platform reported
150
Verified orders
[FACT] Shopify UTM cross-reference
Platform numbers checked against your actual orders.
Google Ads reported 197 conversions over 30 days. Cross-referencing against first-party order data revealed 150 verified purchases — a 24% over-count from blending paid and organic Shopping clicks.
Auto import overwriting manual optimizations
Google-to-MS import tool from previous agency, silently resetting device modifiers and re-enabling paused keywords since Feb 2026.
Automation Detection
The audit uncovered an invisible automation no one knew existed.
A business switching agencies found their optimizations reversing within days. Comparing their Changelog against the actual platform state revealed changes being undone with no record of who made them. The audit traced it to an auto-import tool left over from the previous agency, silently overwriting every manual fix for weeks.
Questions worth asking.
The Bayesian statistical core grades every recommendation, not a generic AI prompt. Each one is gated by a conversion threshold: below it, the audit holds the finding for monitoring instead of forcing a recommendation. Above it, the audit names a credible alternative and the threshold that would shift the call in the next audit. Every claim carries an evidence tag traceable back to your data.
Trellis connects to your ad platforms via OAuth: the same authorization Google, Microsoft, and Meta use. The connection is read-only. Trellis pulls performance data for analysis but never modifies bids, budgets, or campaign settings on your account. Your business economics (margins, targets, COGS) are encrypted at rest and isolated per workspace. No data is shared between clients.
Three things your ad platform’s reporting won’t surface: whether campaigns are profitable after your COGS (not just on revenue), whether platform-reported conversions match the orders your store actually recorded, and which configuration changes caused performance shifts. A campaign showing 3× ROAS may be operating near break-even once your gross margin lands. Trellis catches that; the platform UI does not.
Connect your ad platforms via OAuth in under two minutes. Enter your blended gross margin: one number. Run your first audit. Most accounts see their first scored report within minutes of signing up. No CSV uploads, no technical configuration. The first report tells you how your campaigns are really doing.
Owner-operators, in-house marketers, and agencies running multiple client accounts. The methodology scales to any of them. The same evidence standards and confidence gating apply whether you’re running one campaign or fifty.
Each Catalyst Audit reads your Changelog of the changes you’ve recorded on the ad platforms since the last audit, then measures their actual effect on margin and conversion. After three audits, the model projects budget impact. After six, it runs full inference calibrated to your account, not to a generic benchmark. Each audit gets more specific to how your account actually responds.