Proposal for The RTA Store · Prepared by Craft Digital
Paid Media Infrastructure Modernization
Rebuilding the foundation for scalable, measurable, and AI-ready advertising.
A 30 to 60 day infrastructure build across Google and Microsoft, followed by a six-month strategic advisory engagement that hands the reins to your team.
01 · Executive Summary
Modernizing the infrastructure that powers paid media.
The RTA Store is investing significant resources into paid media across Google and Microsoft. A recent audit identified critical opportunities across attribution, feed management, campaign architecture, Shopping, Performance Max, and reporting.
This initiative is designed to modernize the underlying infrastructure that supports advertising performance. Rather than simply optimizing campaigns, this engagement focuses on rebuilding measurement, feed architecture, account structure, audience strategy, and reporting systems to create a scalable framework that the internal team can confidently manage and grow.
This is a build engagement followed by structured enablement. Craft Digital builds the foundation, proves the data, documents the playbooks, and trains the internal team to own it, supported by a six-month strategic advisory relationship.
Key Outcomes
- Trusted attribution and conversion tracking
- Reduced wasted spend
- Better Shopping and Performance Max performance
- Stronger segmentation and bidding controls
- Documented operating procedures
- Team enablement and training
- AI advertising readiness
02 · Current State Findings
Specific, addressable issues across measurement, structure, and feed.
Each is a lever in this build. The numbers below are the most consequential.
Measurement
- Google Ads revenue mismatch: ~$800K March Purchase value reported in Google Ads vs ~$0 in GA4
- Attribution concerns and unverified attribution model
- Missing Enhanced Conversions
- Unverified purchase tracking
Campaign Structure
- Brand campaigns competing against each other on the same branded terms
- Lack of negative keyword governance
- Weak segmentation
- Shopping inefficiencies
Merchant Center
- Feed categorization problems, including products mis-mapped and disapprovals
- Product classification errors
- Missing segmentation framework (no custom labels)
Microsoft
- Audience Network waste
- Geographic targeting waste: ~$16K spent outside the U.S. in 2026 due to 'people in, searching for, or viewing' targeting
- Missing Shopping implementation
Performance
- Shopping ROAS at 1.44 lifetime and 0.77 trailing-30, currently losing money
- Performance Max readiness issues
- Limited audience utilization: conversion-based audiences and Customer Match unused
03 · Project Overview
Eight stages, three hard dependencies, delivered across six to seven weeks.
Stages shown in relative weeks, not calendar dates. The roadmap can be re-sequenced against a hard deadline such as Q4 or Black Friday.
- ContinuousBrand campaign stays live throughout
- Stage 0 · Week 1Discovery & Stabilization
Foundational access plus short triage to stop active waste.
- Stage 1 · Weeks 1-2GateConversion Tracking Remediation
The gate. Everything value-based downstream depends on this.
- Stage 2 · Weeks 2-3Merchant Center & Feed Architecture
Custom labels and feed cleanup that gate Shopping and PMax.
- Stage 3 · Weeks 2-4Search Rebuild
Google and Microsoft built fresh in parallel, offline in Editor.
- Stage 4 · Weeks 3-4Shopping Restructure
Break apart the 51K-product structure into segmented campaigns.
- Stage 5 · Weeks 3-6Long polePerformance Max
Longest workstream. Build shells early, launch after Search and Shopping are stable.
- Stage 6 · Weeks 5-6Launch & Validation
Sequenced launch with a pre-launch QA gate and client sign-off.
- Stage 7 · Week 6+Handoff & Enablement
Documentation, training, and the start of the advisory cadence.
Critical Path & Dependencies
- Conversion tracking is the gate. Smart bidding (tCPA/tROAS), Shopping ROAS targets, and PMax all inherit whatever is wrong with the current import. Nothing value-based launches until the signal is trusted.
- Feed labels gate Shopping and PMax. Segmenting by margin, product type, price, and category requires custom labels to exist first.
- Search runs in parallel because it is built offline in Editor. Its launch and bidding still wait for tracking.
- The retained brand campaign stays live the entire time, so the account never goes dark.
- PMax is the long pole. Its assembly runs longer than any other workstream, so it starts early as disabled shells even though it enables last.
Launch Sequence
- 1Tracking
- 2Search
- 3Shopping
- 4PMax
With the retained brand campaign running throughout.
04 · Stage 0
Discovery & Stabilization
Foundational access plus a short triage to stop active waste while the real build proceeds.
Access
Systems & Data
- Google Ads, Microsoft Ads, Google Merchant Center, Microsoft Merchant Center
- GA4 and Google Tag Manager
- Magento and the Magento feed source
- Offline conversion source and data table
Activities
Discovery work
- Conversion import investigation
- Revenue reconciliation review
- Attribution review
- Feed source discovery and Magento feed analysis
- Merchant Center diagnostics
- SKU review across ~51,700 products
- Engage the dedicated Google Ads and Microsoft reps and brief them on the action plan
Parallel
Quick Wins (low-risk)
- Disapproval cleanup (mis-categorized products clearing weapons/guns disapprovals)
- Geographic targeting corrections to 'people in' the U.S.
- Audience Network cleanup (placement and website exclusions on Microsoft)
- Starter account-level negative keyword implementation (Google + Microsoft)
- Interim Shopping protection: ROAS floor and/or pause products with $1K+ spend and 0 conversions
Deliverable
Confirmed access, findings report, feed assessment, and finalized rollout schedule.
05 · Stage 1
Conversion Tracking Remediation
Objectives
Tracking & Attribution
- Google Tag implementation
- Purchase tracking validation
- Enhanced Conversions
- Attribution reconciliation
- Revenue verification across Google Ads, GA4, and Magento back-end
- Deduplication validation
- Microsoft UET validation
- Conversion-based audience creation
- Customer Match setup
Exit Criteria
Verified revenue and conversion tracking across Google Ads, Microsoft Ads, GA4, and Magento. This is a go/no-go gate for Stages 4 and 5. Magento dev turnaround is the main schedule risk.
06 · Stage 2
Merchant Center & Feed Architecture
Feed Architecture Flow
Custom Labels
Segmentation backbone
- Margin Tier
- Product Type
- Price Band
- Category
- Best Sellers
- High Margin Products
Feed Optimization
Quality fields
- Titles
- GTINs
- Product Types
- Attributes
- Availability
- Images
Additional
Sync & monitoring
- Supplemental feeds for overrides and enrichment
- Categorization fixes resolving remaining Stage 0 issues at the feed level
- Microsoft Merchant Center synchronization (mirror the feed)
- Feed monitoring framework documented against the Magento generation method
07 · Stage 3
Search Rebuild
Campaign Hierarchy
Google Ads
Rebuild plan
- Brand campaign consolidation: retain one historic campaign as the brand campaign, negatives so it cleanly owns brand traffic
- Non-brand campaign rebuild, net-new, themed by product/category, not match type
- RSA creation (net-new, no in-place editing so historical data stays accurate)
- Asset rebuild: sitelinks, structured snippets, callouts, updated logo
- Negative keyword framework with brand/non-brand separation via shared negative lists
- DSA evaluation as an optional catalog-coverage layer
Microsoft Ads
Mirror & repair
- Structural alignment: mirror Google, import then de-duplicate and theme
- Retargeting implementation: true RLSA with target-and-bid, replacing bid-modifier-only audiences
- Geographic targeting correction to 'people in'
- Staged launch once tracking is live
08 · Stage 4
Shopping Restructure
Before
Undifferentiated mass
- ONE Shopping campaign
- ONE ad group
- 51,713 products
- No subdivisions
- No negatives
After
Segmented, tiered, controlled
- Multiple campaigns by category
- Margin-tier segmentation
- Priority tiers (high/med/low)
- Campaign-level negatives
- ROAS floors applied
Objectives
Restructure plan
- Break apart the 51K+ product structure
- Segment by category
- Segment by margin
- Priority tiers: high / medium / low
- Isolate high performers vs the long tail
- Shopping negatives at the campaign level across all Shopping campaigns
- Apply ROAS targets (>=2.0 floor where data supports) once tracking is verified
- Microsoft Shopping deployment, net-new, since none currently exists
- Document a baseline before changes and phase them to avoid a learning-period shock
09 · Stage 5
Performance Max
PMax Workflow
Workstreams
What gets built
- Campaign shell creation: disabled shells first, multiple campaigns/asset groups by category given inventory size
- Asset group development (images, video, logos, extensive text) - the long pole; start early
- Audience signals using conversion-based audiences from Stage 1
- Retargeting integration: evaluate a dedicated retargeting/RLSA campaign given account size
- Brand exclusions: keep brand traffic in the brand search campaign
- Account-level negative keyword framework
- tROAS configuration aligned to verified conversion values
- Shopping coexistence planning: PMax takes serving priority over Standard Shopping for overlapping products. Decide the coexistence model before enabling.
- Controlled activation: enable after Search and Shopping are stable
10 · Stage 6
Launch & Validation
Launch Sequence
- 1Tracking
- 2Search
- 3Shopping
- 4PMax
Validation
Pre-launch checklist
- Conversion verification
- Revenue verification
- Shopping validation
- PMax validation (not over-cannibalizing Standard Shopping)
- Audience validation
- Geographic validation
- Feed validation (no resurfaced disapprovals)
Supporting Notes
- Wind down legacy structures as new campaigns absorb traffic, no hard cutovers
- Phase budget reallocation to protect learning periods
- De-risk high-impact changes via Drafts & Experiments
- Require a pre-launch QA pass and client sign-off gate
- Stand up baseline reporting dashboards
11 · Stage 7
Handoff & Enablement
Knowledge Transfer Flow
Documentation
SOPs delivered
- N-gram analysis process
- Negative keyword governance
- PMax optimization
- Shopping optimization
- Search testing framework (net-new ad testing, not in-place edits)
- Bidding structure guidance
- Conversion health monitoring
- Budget management framework
- Feed management SOPs (ownership transfers to the client team)
Ownership
Recurring consulting cadence is defined here. Demand Gen / YouTube flagged as a future phase.
12 · AI Advertising Readiness
Preparing for the next generation of advertising.
The advertising landscape is shifting toward AI-native channels. OpenAI has launched a self-serve Ads Manager for ChatGPT (rolling out in the U.S. in 2026), with cost-per-click bidding, new measurement tools, and sponsored placements at the bottom of relevant ChatGPT answers. As an e-commerce brand with a large, structured catalog, The RTA Store is well positioned to be an early mover -- but only if the underlying feed and structured data are ready. This engagement builds that readiness in.
OpenAI / ChatGPT Readiness
Early-mover positioning
- Position The RTA Store for emerging AI advertising ecosystems (OpenAI Ads Manager and conversational-commerce placements)
- Conversational commerce readiness
- Product feed readiness: the same clean, labeled feed built in Stage 2 is the foundation AI ad channels will consume
- Structured data readiness
- AI search visibility
- Early-mover positioning and a test-and-learn plan for ChatGPT ad placements as access expands
Sources: OpenAI's ads announcements and 2026 coverage of the ChatGPT Ads Manager rollout. Specific MCP architecture, hosting model, and connector list are finalized during technical discovery at kickoff.
13 · AI-Powered Ad Operations Layer
An agentic operations layer inside your team's workflow.
A committed deliverable of this engagement -- not a stretch goal. We build the tooling and routines; your team runs them; we consult on the review.
Beyond the audit fixes and campaign restructure, Craft Digital stands up an agentic operations layer inside your team's own workflow. The daily hygiene work that keeps a $250K/month account healthy gets done every day, automatically -- and your team's time shifts to the decisions that actually move MER. We are not the agency and we are not replacing Andrew or Sazid. We build the tooling and routines, your team runs them, and we consult on the review.
This is the section that turns the audit into ongoing infrastructure value, not just a consulting engagement -- and the direct answer to the question from the June 11 call about why an in-house team would also want managed-service work.
1 · Platform integrations
Google Ads + Microsoft Advertising
- An MCP (Model Context Protocol) server that brokers secure connections between your ad/analytics platforms and an AI assistant (Claude, ChatGPT, or the team's tool of choice)
- Built on the live Claude + Google Ads connection Craft Digital already proved during the audit -- hardened, then extended to Microsoft Advertising so both paid-search channels are queryable and manageable from one place
- Read access powers reporting and monitoring from day one; write actions (negative keywords, budget adjustments, asset swaps) run behind approval gates so a human signs off before anything changes
- Hosted on infrastructure The RTA Store owns; credentials and data stay in your environment
Connected data sources
One queryable layer
- Google Ads
- Microsoft Ads
- Google Merchant Center
- Microsoft Merchant Center
- GA4
- Magento back-end revenue (read-only) for reconciliation
2 · Skills your team can run on demand
Slash commands, packaged from the audit findings
Full sanity check across structure, settings drift, missing negatives, disapproved ads, and stale assets (like the outdated logos flagged in the audit). Same idea as a lint check before code ships -- nothing runs broken for weeks unnoticed.
Sweeps recent search-term reports across Google and Microsoft, proposes negative keyword candidates with spend wasted per term, queued for one-click approval.
Month-to-date pacing by campaign vs. plan. Flags over- and under-spenders and ties back to the MER target.
Audits live ad copy and assets against the current brand and sale calendar -- including compliance flags on sale-extension language.
Performance summary in your terms (leads in Zoho, sales by UTM in Magento) -- not just platform-reported conversions.
3 · Routines on a set cadence
The same skills, running automatically
The agent does the brute-force pass; the output is always "here is what I found, here is what was handled, here is what needs a human."
- Budget pacing and anomaly check
- Disapproved-ad and broken-landing-page alerts
- Search-term sweep with negative-keyword candidates ready for approval
- Account performance recap against MER
- Bid and quality outliers
- Asset freshness check
- Structure-drift review against the rebuilt campaign architecture
- Creative-fatigue flags to feed the design team's refresh loop
What this frees your team to do
Negative keywords, pacing checks, asset hygiene, and report pulls are necessary work -- but they are not what you hired senior people for. With the routine layer handling that floor, Andrew and Sazid's time goes to the high-value side of advertising management: deciding where reallocated test budget goes (Reddit, CTV, Pinterest), running structured channel tests, tightening the creative refresh cadence with your designers, and evolving the attribution model. Craft Digital stays involved as the guardrails -- reviewing what the routines surface, pressure-testing decisions, and extending the skill set as the account matures.
Deliverables
What Craft Digital ships
- Provisioned MCP server connected to all in-scope data sources, with approval-gated write actions on Google and Microsoft
- Skill library (/account-health, /search-terms, /budget-pacing, /creative-check, /weekly-recap) installed in your team's Claude workspace
- Scheduled routines (daily / twice-weekly / weekly / monthly) running automatically, with results delivered to the team
- Documentation: how it works, how to connect a new client app, how to add or rotate data sources
- Training sessions for the internal team and stakeholders on how to use it well
- Operating SOP so the team can extend it (new skills, new data sources, new routines) after handoff
Ownership & Handoff
Ownership transfers to The RTA Store at the end of the build. Craft Digital supports the operations layer during the 6-month advisory engagement -- helping the team add skills, refine routines, and roll it out to additional stakeholders -- but the infrastructure, credentials, and data stay with you from day one.
Example natural-language queries the team will run
What cabinet categories generated the highest ROAS this month?
Top 3 by ROAS: Shaker White (4.2x), Grey Stone (3.8x), Espresso Maple (3.1x). Combined spend $42K · revenue $151K.
Which products have over $1,000 in spend and zero conversions?
27 SKUs flagged. Combined spend $34,820 with zero attributed conversions in the trailing 30 days. Recommend pause + feed review.
Compare branded vs non-branded search performance week over week.
Branded CPC +8% WoW · Conversions +14%. Non-branded CPC -3% · Conversions -6%. Branded efficiency improving; non-brand needs creative test.
14 · Strategic Advisory Engagement
$5,000 / month -- 6-month minimum.
Once the infrastructure is built, Craft Digital stays on as a strategic advisor -- monitoring, reporting, and guiding the internal team while they take the reins.
Monthly Strategy Sessions
- Performance review
- Attribution review
- Feed review
- Campaign review
- Budget review
- Testing roadmap
Weekly Reporting
- KPI reporting
- Executive summaries
- Risks
- Opportunities
- Recommendations
Ongoing Consulting
- Team support
- Platform updates
- Optimization guidance
- New opportunities
- Strategic planning
Cadence
During the build, Craft Digital meets with the team biweekly so they can see what is being built. After launch, the team moves to monthly (or biweekly) monitoring reviews, with weekly reports delivered throughout the six-month engagement.
15 · Expected Outcomes
What this engagement leaves behind.
Cleaner Data
Verified attribution and trusted revenue signals reconciled across Google Ads, Microsoft, GA4, and Magento.
Reduced Waste
Improved targeting, exclusions, and campaign governance. Branded and non-branded separated so the account stops bidding against itself.
Better Segmentation
Category, margin, and performance-based structures instead of one undifferentiated 51K-product campaign.
Improved Automation
Reliable smart bidding and Performance Max signals operating on real value.
Internal Enablement
Documented SOPs and a trained team that can run and optimize the account.
AI Readiness
Prepared for future advertising channels and AI-driven opportunities, with an MCP operations layer in place.
16 · Investment
The decision.
Implementation Phase
Infrastructure Build
One-time investment
The full 30 to 60 day infrastructure build (Stages 0 through 7): discovery, conversion tracking remediation, feed architecture, search rebuild, Shopping restructure, Performance Max, launch, handoff, and AI/MCP enablement.
Strategic Advisory & Enablement
6-Month Advisory
Per month · 6-month minimum
Monthly strategy sessions, weekly reporting, and ongoing consulting as the internal team takes ownership.
Total Engagement Value
$50,000
$20,000 implementation + $30,000 advisory over six months
This is a build engagement with a structured handoff. Craft Digital builds the foundation, proves the data, and trains your team to own it.