Overview
Summary
Once the site is live and the readiness gate is met, marketing runs as a loop with three steps: plan one test with one to three numbers attached, facilitate it with the copy, briefs and checklists it needs, and analyse what happened against those numbers. The analysis is the first thing the next plan reads, so every cycle starts from evidence rather than opinion.
Paid channels run monthly cycles; local search and partnerships run quarterly. We run the machinery and write the analysis; Paynes answers the phone, quotes, does the work and records outcomes so the numbers mean something.
Section 1
The cycle
Plan, facilitate, analyse, then the next plan. Each step has a fixed shape so cycles can be compared.
One hypothesis in a sentence. One to three KPIs from the menu, each with a target and a named source. Audience, channel, message in the brand voice, assets, schedule, budget, review date. Never more than the business can answer.
Everything the plan needs to run: advert copy and variants, a page change if needed, a posting schedule, a partner briefing, a launch checklist (tracking working, budget cap set, number answered), and a mid-cycle check on spend and tracking.
Actuals against targets from the named sources; whether the hypothesis held; what worked, what did not, and why we think so; three recommendations, each naming the section of the plan it would change.
Nothing else carries state between cycles. The dated folder is the memory, and the previous analysis is the first input to the next plan.
Section 2
Cadence
Monthly where events accumulate quickly, quarterly where they do not, and one analysis a month covering every live cycle.
| Activity | Cycle | Check-in | Why |
|---|---|---|---|
| Paid search and social | Monthly | Weekly: spend, tracking, obvious failures | Enough events to judge; short enough to stop waste. Not judged on less than four weeks or fewer than about thirty contacts, whichever comes first. About 100 a month is a small budget, so early cycles may need the full four weeks. |
| Local search and the Business Profile | Quarterly | Monthly | Rankings and profile changes move slowly. |
| Partnerships (builders, estate agents) and offline | Per activity | At the activity | A mailing, an introduction or a season has its own timetable. |
| Reporting | Monthly | First working day of the month (proposed) | One analysis a month, covering every live cycle. |
Section 3
Who does what
We run the machinery. Paynes answers, quotes, does the work and records what happened.
| Step | We | Paynes |
|---|---|---|
| Plan | Write the plan from the last analysis; propose the hypothesis, budget and message. | Approve the budget and the message; say what the business can take on that month; flag seasonal demand and offers. |
| Facilitate | Produce the adverts, posts, page changes and briefs; set up tracking; launch. | Supply photos of finished jobs, prices and facts; answer every call and form within the agreed time; log enquiries that arrive off the tracked routes. |
| Analyse | Pull the numbers from each source; write the analysis; make three recommendations. | Share job counts and values from the lead record; mark leads won or lost with a reason; read the analysis and make the decisions it asks for. |
Section 4
The KPI menu
Every plan picks its KPIs from this list, so the same thing is measured the same way every cycle. A KPI needs a source; while the source does not exist, the KPI is not available and the plan says so.
| Phase | KPI | Source | Available today |
|---|---|---|---|
| Reach | Impressions, clicks and cost per channel | Google Ads, social ad platform | no accounts to be set up |
| Reach | Search impressions and clicks | Search Console | no verify after go-live |
| Reach | Profile views and calls from the listing | Business Profile insights | no profile status unknown |
| Land | Sessions by source, landing-page views | Analytics (GA4 or PostHog, to be chosen) | no |
| Contact | Phone taps, form sends | Analytics events | no |
| Contact | Calls answered and missed per tracked number | Call tracking (to be chosen) | no |
| Capture | Enquiries logged, share with a known source, response time | Lead record (to be chosen) | no |
| Quote | Quotes sent, quote value, win rate | Lead record | no |
| Execute | Jobs won, revenue, average job value | Lead record and accounts | no |
| Retain | Reviews gained and rating | Google reviews | once the profile is confirmed |
| Efficiency | Cost per contact, cost per quote, cost per job | Ad platform spend joined to the lead record | no |
Section 5
Decision rules
Five rules, applied in the analysis, so the next plan is a consequence rather than a preference.
When a channel's KPI beat target for a full cycle and cost per contact is at or under target. Raise the budget one step per cycle, never more than one step.
When reach is fine but contact rate is low (the message or the page is the problem), or contacts are fine but quotes are low (response time or qualification is the problem). Change one variable per cycle so the hypothesis stays testable.
A channel that misses target two cycles running after one adjustment. Move its budget to the best performer.
When tracking was broken or the data is too thin to judge. Fix the measurement first and repeat the cycle.
Plan calls the business cannot answer. Calls that go unanswered are a failure at Contact, not a success at Reach, and the analysis will say so.
Section 6
How one cycle feeds the next
Each run lives in its own dated folder with three files. The next plan opens the previous analysis first.
The hypothesis and the KPIs.
What was produced to run it.
What happened and three recommendations.
Answers each recommendation: adopt, adapt or decline with a reason.
The KPI menu above is updated in the same run if a plan needs a KPI it does not list. A run is not complete until its analysis exists; an abandoned run gets an analysis that says so and why.
Section 7
A worked example (illustrative, not a plan)
Numbers below are placeholders to show the shape; real targets are set in the first plan from the first two weeks of data.
Plan
Hypothesis: Google Ads call-only adverts for "emergency plumber" and "boiler repair" in Uckfield, Lewes, Tunbridge Wells, Brighton and Haywards Heath produce answered calls at an acceptable cost.
KPIs: answered calls from the tracked number (call tracking); cost per answered call (Google Ads spend divided by those calls); enquiries logged with source "Google Ads call-only" (lead record).
Budget: the monthly amount agreed. Review date: four weeks from launch.
Facilitate
Three advert variants in the brand voice, each leading with a fact (since 1990, Gas Safe registered, 24-hour call-out); a tracked number that forwards to the main line; the launch checklist; a weekly spend and tracking check.
Analyse
Answered calls against target; missed calls and when they happened (if most are out of hours, the recommendation is about answering, not advertising); cost per answered call; how many became quotes.
Three recommendations, for example: change the schedule to daytime only (plan section: schedule); add "Uckfield" to the headline for the best town (message); move part of the budget to boiler installation search adverts for the winter (channels).
Next plan
Reads that analysis, adopts or declines each recommendation, and states the next hypothesis.
Check
Verify
Every KPI in section 4 names a source the pipeline overview lists. Roles in section 3 match the shared agency file. Cadence and decision rules follow the stage's loop model; the only client-specific additions are the budget note and the proposed reporting day. No dated run folder exists because the readiness gate is not met. This page has the same sections as marketing-playbook.md in the same order.
For Paynes
Open questions
- Reporting day: is the first working day of each month right, and who at Paynes receives the analysis?
- What is the agreed response time for calls and forms during opening hours?
- Which month should the first paid cycle start, once the gate is met? Heating demand rises from autumn; renewables enquiries are steadier.
- Is there a maximum number of new jobs a month the team can take on? The plan must not exceed it.