The short version
- Day thirty is an inventory checkpoint. Six checks, each with a written answer and a name against it.
- Roughly 92% of one client's raw sessions were bot, which had quietly distorted every ratio anybody had quoted for years.
- Learn the publish path before you have write access. On one rebuild the obvious deploy command would have gone over the live site.
- Ship in week one. Holding delivery until discovery finishes costs the engagement its earliest and cheapest correction points.
By day thirty a new engagement needs six things true: we can publish, we know what is actually live, the analytics are filtered, the CRM moves contacts past the first stage, every approval has a named owner, and every tool ceiling is written down. Strategy built on an unverified stack inherits all of its errors.
The first month is an inventory pass with delivery running alongside it.
What has to be true by day thirty?
Six checks, each with a written answer and one name against it. None of them are strategy, and all of them decide whether the strategy is worth writing.
The six checks, and what they have turned up on real engagements:
| Check | The question it answers | What it has turned up |
|---|---|---|
| Publish path | How does a change reach production? | A push to a branch. The obvious deploy command would have overwritten the live site. |
| What is live | Which application serves the domain? | The staging URL in the project doc served a different app. Production ran a different framework from the repo. |
| Analytics | How much of this traffic is human? | Roughly 92% of raw sessions were bot, with filtering never switched on. |
| CRM | Does anything move a contact forward? | Roughly 98% of contacts sat in the first lifecycle stage. |
| Gates | Who can stop a launch? | Two approvals nobody had listed: security review and pricing sign-off. |
| Tool ceilings | What is the limit on each tool? | A free email tier capped sends at 300 a day. |
Why does the inventory come before the strategy?
Because every number in a first strategy document is computed off the stack. Where roughly 92% of raw sessions are non-human, the distortion is worst in the top-pages report, since crawlers concentrate on particular paths. A content team reading it commissions more of whatever the bots liked.
The same holds at the other end. Adding traffic to a CRM where almost everything sits in the first lifecycle stage buys names rather than pipeline, because nothing downstream moves anybody forward. That case is audit the CRM before you add traffic.
What happens in week one?
Read access to everything first, confirmed by opening each one: analytics, Search Console, the CRM, ad accounts, the CMS and the repository. Then one trivial visible change followed end to end into production before anything real is touched.
Learn the publish path before you hold write access rather than after. And ask who can stop a launch, a question that turns up approvals held outside marketing which no delivery plan lists.
Who owns what by day thirty?
One named person per item, on both sides. During one rebuild a blog stopped serving. Everybody could see it and it appeared on status lists, and it still stayed down for weeks. The technical fix was not difficult. It sat between two vendors and two internal functions with nobody’s name against it.
Ownership gets recorded where both sides can see it. A shared channel per relationship rather than scattered DMs, because a decision made in a DM does not exist for anybody else, and a portal per client so the record of what shipped is structural.
Temporarily down became permanent at the speed of whoever was willing to keep raising it.
What does the first thirty days produce?
A written stack inventory, a gate list with owners and lead times, reporting that runs on filtered data, and the first weekly deliverables already out.
The cadence starts in week one. A weekly rhythm is worth roughly fifty correction points in a year against twelve for a monthly one, and the earliest ones matter most, because that is when a wrong assumption is cheapest to catch.
The two checks worth running before anything else are verifying what is actually live and what happens when nobody’s name is on it.
Frequently asked questions
What should happen in the first 30 days with a new marketing agency?
An inventory pass, with the first delivered work running alongside it. By day thirty the agency can publish, knows which application is actually live, reports on filtered analytics, has a lifecycle map for the CRM, and holds a written list of every approval gate and its owner.
What access does an agency need during onboarding?
Read access to analytics, Search Console, the CRM, ad accounts, the CMS and the code repository, each confirmed by opening it. Publish access comes after the publish path is documented, because that is the step where an obvious command can overwrite production.
Why do agency engagements go wrong in the first month?
The strategy gets built on numbers nobody checked. An analytics audit on one account found roughly 92% of raw sessions were bot. A CRM audit on another found roughly 98% of contacts stuck in the first lifecycle stage. Both had been true for years.
How do you audit a new client's analytics?
Check the bot share before quoting any ratio. Filtering is frequently never switched on, and the distortion concentrates in the top-pages report because non-human traffic clusters on particular paths. Every ratio built on the raw denominator has to be recomputed once filtering is on.
Should an agency deliver work in the first month or only plan?
Deliver. The inventory pass runs alongside the first deliverables rather than in front of them. A weekly cadence produces roughly fifty points a year where a client can look at real work and correct it, and the first few matter most.
What is a stack inventory?
A written record of how a client's marketing systems actually behave: the publish path, which application serves the domain, whether analytics are filtered, how the CRM moves contacts, who holds each approval, and the hard limit on every tool.
Sources
- Chua Network delivery data across 8 client accounts (internal fact bank)
- Chua Network engagement records, anonymized (internal experience bank)
