GSI

Onboarding

Your first 90 days with Store Intelligence.

A clear, week-by-week plan. Connect and structure your store in weeks 1–2, run your first approved AI work in weeks 3–4, scale across the catalogue in Month 2, and start compounding from Month 3.

No three-month black box. Every phase has concrete deliverables, a defined ask of your time, and a measurable outcome at the end. If something isn't working at the end of a phase, you know before the next one starts.

01

Month 1 · Weeks 1–2

Foundation — connect, sync, and structure

Wire the system into your store, your analytics, and your search data. Build the structured knowledge layer that everything else depends on.

What we deliver

  • Connect your store (Shopify, Neto, BigCommerce, or hybrid) via OAuth — no platform migration required.
  • Connect Google Analytics 4 and Google Search Console; sync 12+ months of historical data.
  • Run URL reconciliation — map every current URL to its history, redirect chains, and traffic lineage. Identifies hidden traffic loss from URL changes.
  • Capture your store context document with you: brand voice, audience descriptors, tone-by-category, regulated language, words to avoid.
  • Propose an initial MECE category structure based on your catalogue — review and refine with you in one working session.
  • Run a baseline catalogue audit. Establish current health scores per category, missing-attribute counts, content quality benchmarks.

What you do

  • Grant platform access (OAuth flows take ~15 minutes total).
  • Join one 60–90 minute kickoff call to walk us through your store, voice, and what good looks like.
  • Review the proposed MECE structure (~1 hour) — give feedback on category names and boundaries.

Outcome

By end of week 2, you have a dashboard with your full catalogue structured, health-scored, and historically attributed. You can already see what your store has been hiding.

02

Month 1 · Weeks 3–4

First tasks live — initial AI work, fully approved

Run the first batches of AI-assisted improvements on real products. Establish the approval rhythm. Score and refine prompts based on your feedback before scaling.

What we deliver

  • Configure the first two AI tasks together: SEO page titles and meta descriptions. We pick these because they're high-impact, low-blast-radius, and easy to evaluate.
  • Build per-category prompts that inherit from your store context — different voice for tents vs. cookware, but consistent across each category.
  • Run the first batch through the approval queue (typically 50–200 products) with confidence scoring and diff preview on every change.
  • Score every output together — what landed, what missed, why. Refine prompts to v2 based on what you flag.
  • Run a second batch with refined prompts. Lock in quality benchmarks before we scale.
  • Publish approved changes to your live store; everything is recorded and one-click reversible.

What you do

  • Review the first batches and approve / reject / edit (~2–3 hours per week for the first fortnight).
  • Tell us what's off — tone, accuracy, brand voice — so we can tune prompts before scaling.

Outcome

By end of week 4, you have your first measurable improvements live, two scored and validated prompt configurations ready to scale, and a working rhythm for approving AI work going forward.

03

Month 2

Scale — broader coverage, deeper granularity

Expand AI assistance from the first two fields to broader catalogue work. Tackle data quality issues at scale. Build per-category granularity so the AI gets sharper the more it runs.

What we deliver

  • Extend AI tasks to product descriptions, category descriptions, image alt text, and structured-data attributes.
  • Run catalogue-wide audits — surface missing attributes, thin content, conflicting categories, broken meta lengths, keyword cannibalisation.
  • Build per-category prompts at finer granularity, using Month 1 learnings on what tone and structure work where.
  • Move to larger batch cycles — thousands of products per overnight run, every change still queued for approval.
  • Begin weekly digest emails: what changed, what improved, what's queued.
  • Run your first monthly growth review at the end of Month 2 — measured movement, prioritised next round.

What you do

  • Weekly approval session (~1 hour) — your team takes more of the approval load as confidence grows.
  • One 30-minute monthly review to look at numbers and choose next priorities.

Outcome

By end of Month 2, the AI is running across thousands of SKUs, tuned per category, with your team approving in batches. You have your first month-over-month numbers showing what moved and what didn't.

04

Month 3 and ongoing

Compound — research, hypothesise, test, measure

Move from one-time improvement to continuous compounding. Every quarter is a research cycle: form hypotheses, test them at category scale, measure outcomes, lock in winners.

What we deliver

  • First quarterly growth review at end of Month 3 — full numbers, root-cause analysis, prioritised next quarter.
  • Research mode: dig into what's working and what isn't. Where are the next wins hiding?
  • Form testable hypotheses with you ("what if we restructured the X category?", "what if brand voice for Y were tightened?").
  • Build A/B prompt tests — compare prompt versions head-to-head on subsets of your catalogue and measure outcomes.
  • Run tests at category scale. Measure on rankings, CTR, conversion, revenue — not just published-change counts.
  • Lock in winners, retire losers. Each test result calibrates the next round; your store gets sharper every quarter.
  • Ongoing audit cycles, monthly digests, quarterly growth reviews. Your team owns more of the day-to-day; we step back to research, strategy, and hard cases.

What you do

  • Quarterly strategy session (~2 hours).
  • Day-to-day approvals run through your team with us in the background for hard cases.

Outcome

From Month 3 onward, the store improves measurably every quarter. Every change has provenance, every test has a result, and the AI gets sharper with every cycle — because every approval, rejection, and revert calibrates what comes next.

Responsibility split

Who does what.

The whole point of a managed service is that we carry the operational weight while you keep control of the decisions. Here's exactly where the line sits across the engagement.

AreaYouStore Intelligence

Platform access & connections

Provide OAuth access (~15 min total)

Wire everything in, validate the syncs, troubleshoot

Store context & voice

1× kickoff call + answer brand-voice questions

Capture, structure, version. Update as you grow

MECE structure

Review proposals + 1 hour of feedback

Propose the structure, refine, version

AI prompt configuration

Tell us what's off when you see it

Build, refine, score, A/B test, retire what doesn't work

Reviewing AI proposals

Approve / reject / edit — heavy at first, lighter as confidence grows

Queue them, score them, surface the highest-impact first

Publishing & rollback

Click approve. That's it.

Publish to your store, record the change, manage the rollback safety net

Reporting & reviews

Show up to the monthly + quarterly reviews

Generate the digests, prepare the reviews, propose next priorities

What if…

The questions we get on every onboarding call.

What if my product data is messy?

It almost always is. Weeks 1–2 surface exactly what's messy and rank it by impact. You don't fix everything; you fix what costs you money, in order. The audit is part of the value, not a prerequisite.

What if I want to slow down?

You can. We work in fortnight increments. If a sale, a holiday, or a launch needs your attention, we pause the publish queue and pick back up. Approvals never queue indefinitely — we re-score against current state when you return.

What about Black Friday, EOFY, peak season?

We freeze publishing windows around your peak periods. Audits and proposals continue in the background, batched up for after peak. Nothing publishes to live during freeze windows.

What if my team is small?

Most of our customers run lean. The whole point of approval queues + batch review is that it scales to one reviewer. We tune prompt quality up so approval rates stay above 80% and review time stays under an hour per week.

What if I want to bring it in-house later?

Good. That's the goal. Months 1–2 are heavy-touch from us; Month 3 onward your team owns the day-to-day with us in the background. The platform stays consistent regardless of who's driving.

What does it cost during onboarding?

The founding partner rollout is a one-off cost for the Month 1–2 implementation. Month 3 onward is a monthly service relationship. See the pricing on the home page or ask on the call.

Ready to map this to your store?

20-minute call. We'll walk through where your catalogue sits today, where you'd be in 30 days, and whether this engagement makes sense for your store. No deck, no pitch — just an honest read on whether we're the right fit.