Internal strategy outline · Hudson & Nine
Kim's cap is $1,000 a month including our fee. That number decides almost everything below. It rules out Google Ads as an acquisition channel and rules it in as a cheap, tightly controlled measurement instrument. This is what we would build, in what order, and how we would know whether to keep it.
01
One sentence, because everything else follows from it.
This account has never been measurable, so it has never had a fair test. The strategy is not to spend better. It is to make the account able to answer a question, then ask it once, cheaply, and act on the answer.
A conversion action called Qualified Lead (Offline) sits in the account, marked Primary, and has never once fired. Google has spent eighteen months optimising toward form submissions because that is the only signal anyone gave it. Kim's complaint is that leads do not convert. That is the same fact stated from the other end.
Georgia Dilar's own handover chain confirms it independently: the click ID and timestamp capture work, but the step that moves qualified leads into the Ads-formatted sheet, and the import that returns them to Google, were never built. The plumbing is roughly eighty percent done and empty.
02
Before designing anything, the arithmetic has to be said out loud.
The verified year to date cost per lead in the two ad groups that matter is $169 and $182. At $750 of media that is roughly four inquiries a month and twelve to thirteen across ninety days.
Twelve is enough to see whether tight corporate targeting pulls genuinely corporate inquiries. It is nowhere near enough to prove a return statistically. Anyone presenting a ninety day test as an ROI answer at this budget is overselling it.
Break even requires a true lead value of at least $151.89. The value currently in the account is $1,500 spread across 42 conversions, about $36 each, which is a placeholder someone typed. Until Kim supplies average booking value, close rate and margin by service line, nobody can say whether $182 an inquiry is a bargain or a loss.
One further constraint: the Search campaign currently runs at $45 a day, about $1,350 a month, and already forfeits 43.95% of available impressions to budget. Cutting media to $750 is a 44% reduction. That is only survivable because the strategy shrinks the auction pool at the same time. Narrow the targeting and the smaller budget covers a larger share of a smaller market. Keep the current sprawl and $750 buys nothing but thinner coverage of the same waste.
03
No media changes. Nothing here costs ad spend. All of it is prerequisite.
It has run without a manager since 31 August and is still spending roughly $45 a day toward a signal nobody is reading. Every day of deliberation costs about $45. Pausing is reversible and free.
Build the step that moves qualified leads from the raw sheet into the Ads-formatted sheet, then connect the import so Google receives them. This is the single highest leverage change available anywhere in the engagement, and most of it already exists.
The import is worthless if the definition is vague. It needs a rule Kim can apply in ten seconds per inquiry: budget band, event type, timeline, and whether it reached a conversation. This is a client task we cannot do for her, and it is the real dependency in the whole plan.
Nine actions are enabled: one Active, seven Misconfigured, one Awaiting. Separately, the cheq invalid click action inflates any all conversions view and must be excluded from every report. Anyone reading the headline number cold will report 55 leads. There are not 55 leads.
A third party click fraud tool still holds write access and was active on 3 September. Nobody is managing it. It should not have write permission on an account we are about to rebuild.
AI Max is on, matching beyond the keyword list. $4,024.58 of $7,088.12, 56.8% of lifetime spend, and 26 of 45 conversions sit behind an "Other search terms" bucket Google will not itemise. A test whose inputs cannot be named is not a test.
04
One campaign. Two ad groups. Deliberately smaller than what exists.
Seven service line ad groups already exist. At $750 a month, seven ad groups means none of them gathers enough data to be read. The verified evidence says the choice is already made for us.
| Ad group | Year to date spend | Conversions | Verdict |
|---|---|---|---|
| Brand Activation | part of $1,585.59 | 9 of 10 between them | Keep. Carries the account. |
| Corporate Events | part of $1,585.59 | 9 of 10 between them | Keep. Carries the account. |
| Corporate Gifting | $147 | 0 | Pause. Kim also doubts the margin. |
| Non-Profit Gala | $20.30 all year | 0 | See section 06. Not a fair reading. |
| Dynamic search ads | $660.98 lifetime | 1 | Already paused. Keep it that way. |
| Interior Design | $355.15 lifetime | 0 | Already paused. Not their business. |
Brand Activation and Corporate Events took $1,585.59 of the $2,127.49 Google can itemise year to date and produced 9 of the 10 conversions. Kim proposed concentrating there before seeing this table. Her instinct and the data agree.
The enabled keyword set today is 359 broad, 139 phrase, 16 exact. Broad is the majority and it is the reason nothing is legible. The rebuild inverts that:
Caveat worth stating internally: with roughly 100 clicks a month, keyword level decisions will not reach significance. We tested the existing keyword rankings and best versus worst came out at p = 0.20, statistically indistinguishable. Match types are for visibility and control, not for keyword optimisation. We should not pretend otherwise in the reporting.
The existing negative list is extensive but built reactively and in exact match, so it blocks terms only after they have cost money. Eight competitor terms flagged Excluded had already cost $87.75 before being blocked, and freeman tradeshows is flagged Added, meaning a competitor brand was bought as a keyword for $16.00. One click on maritz cost $55.75 and was never on the list at all.
Total identifiable competitor brand spend is $159.50, 7.5% of named spend. Use the single $55.75 click as the illustration. Never quote an aggregate percentage, and never the withdrawn 68% figure.
05
A corporate specific page, which Kim asked for, but built after phase zero and not before. It needs to do three things: state the category and the calibre of client plainly, carry named proof rather than a logo wall, and put a budget qualifying field in the form.
That budget field is the mechanism that solves the small budget inquiry problem at the source, and it is also what feeds the qualified lead definition.
Start on manual CPC or maximise clicks with a bid cap. Smart bidding needs roughly 30 conversions in 30 days to learn. This account will produce four a month. Handing bidding to an algorithm at that volume is how the last eighteen months happened.
Revisit target CPA only once the offline import has been feeding qualified leads back for a full quarter, and only if volume supports it. It probably will not.
Performance Max stays off. It loses more than 90% of its impression share to rank, not budget, so additional money there achieves close to nothing. Its apparent $36.06 cost per conversion is flattered by at least one verified brand search, and brand searches are traffic Hudson & Nine would have received for free.
06
Kim wants nonprofit paused because it attracts small budget inquiries. The inquiries are real. The conclusion does not follow.
Non Profit Event ad group produced the cheapest leads in the entire account at $41.64 each.Non-Profit Gala ad group received $20.30 for the whole year. It has never been funded, so its zero conversions are not evidence of anything.The right instrument is the budget qualifying field on the form, not category deletion. Nonprofit galas are not inherently small. If Kim overrules this, we follow her, but it should be recorded that we advised otherwise and why.
Caveat: the $41.64 figure sits inside a paused campaign from the earlier era, when clicks cost $4.19 against $7.46 now. Cheaper clicks explain part of that advantage and possibly most of it. It is suggestive, not proof.
07
Weeks 0 to 2 · no media
Campaign paused. Import built and tested with real submissions. Qualified definition agreed with Kim in writing. Conversion actions repaired. CHEQ access revoked. AI Max off. Nothing launches until a test lead has travelled the full loop from the form back into Google Ads.
Weeks 3 to 4 · $750 a month begins
One campaign, two ad groups, exact match carrying the budget, full negative list live, corporate landing page receiving the traffic. Manual bidding. Weekly search term review, harvesting into exact and adding negatives.
Weeks 5 to 8
At four inquiries a month the temptation to react to every data point is the main risk. Changes limited to negatives and obvious waste. No structural changes, because there will not be enough volume to justify one.
Weeks 9 to 12
The question is not how many forms were filled in. It is how many inquiries Kim marked qualified, and whether any reached a conversation.
| Outcome after 90 days | Decision |
|---|---|
| 3 or more inquiries marked qualified, and at least 1 reaches proposal stage | Continue. Revisit budget with real numbers. |
| 12 inquiries, none qualified as corporate | Stop. Redirect the budget to Track 1. |
| Quality is visible but volume is too thin to read | Extend once, and only if Kim chooses to raise media. Not on our recommendation alone. |
Writing the stop condition down before launch is the part that matters. It is what turns this from another open ended retainer into a test, and it is what Kim explicitly asked for.
08
In Hudson & Nine's own analytics, visitors arriving from AI assistants engage at 70.97% for 51 seconds. Direct traffic, which is 60.58% of all sessions, engages at 21.62% for 12 seconds. The smallest channel is behaving the best by a wide margin.
For a business people hire on judgment, creative work and demonstrated results, being the studio an AI assistant recommends is a better structural fit than bidding against agencies with far larger budgets. That is the argument for Track 1. Google Ads at $1,000 a month is the small controlled experiment running alongside it, not the main effort. Saying that plainly costs us a little revenue and buys the credibility that keeps the rest.