Methodology

Nothing is bought, built or switched on until I know what is already there. The audit covers the ad accounts and how they are structured, what has been spent and against what, the state of the tracking, the site the traffic lands on, whatever search visibility already exists, and what competitors are visibly doing. Just as important is the commercial half: what you sell, what a customer is worth, which services actually carry margin, how an enquiry is handled once it arrives, and what has already been tried and abandoned. Most expensive marketing mistakes are made at this point, by skipping it — a campaign launched onto broken tracking, budget aimed at a service the business does not want more of, or a rebuild of the one thing that was working. The audit ends as a written picture of where you stand and what is holding you back. It is deliberately useful whether or not the work goes further; you should be able to hand it to anyone.

Then the plan narrows, which is the part most proposals skip. The instinct with a new engagement is to be visible everywhere at once — search, paid social, email, a blog, a newsletter — and the result is five channels each running at a fifth of the attention it needs, none of them accumulating enough data to be judged. So this stage is mostly a decision about what not to do. Channels are chosen against three things: where your buyers actually are at the point they decide, what the budget can realistically feed to the level where a platform can learn, and what the business can support operationally — there is no sense generating enquiries at midnight if nobody can answer before Tuesday. For most businesses that means one or two channels to begin with, with a third added only once the first is producing something worth scaling. The channels that were not chosen get written down with the reason, so the decision can be revisited when the constraint changes rather than reopened from scratch every month.

Metrics chosen after a campaign has run tend to flatter whoever chose them. So before any budget goes live we agree, in writing, which numbers define success, where each one is sourced, who owns the report, how often it is reviewed, and — the clause usually left out — what result would make us change course. That last part is what turns a dashboard into a decision-making tool instead of a monthly performance. This is also where a qualified enquiry gets defined. Lead is the most abused word in this industry precisely because it is elastic enough to hit any target; if the sales side and the marketing side have not agreed what counts before the spending starts, the reporting afterwards becomes an argument rather than a fact. Alongside the definitions, the tracking itself gets fixed: conversion events firing correctly and only once, server-side tagging where browser events are being dropped, consistent UTM conventions, and an end-to-end check that an enquiry can genuinely be traced back to the campaign that caused it.

Only now does anything get built. Campaigns are structured around the way you actually sell rather than the way the platform's default wizard suggests, and the first launch is deliberately narrow: a small number of campaigns, a small number of creative angles, and landing pages built for those specifically rather than one general contact page asked to serve everybody. A narrow launch is not caution for its own sake — it is what makes the results readable. Ten variables changed at once produce a number nobody can explain afterwards. Where the work is search rather than paid, the same principle applies in a different order: technical fixes ship first, then the priority pages are rewritten against real search intent, then content is published against the mapped keywords. Anything touching a page that already performs goes out carefully and is watched rather than pushed in one move. Expect part of the first set to be wrong; that is what the early budget is for, and being wrong in week four on a small budget is far cheaper than being wrong in month three on a large one.

Weekly is the rhythm the work actually runs at. Each week: what the data has changed its mind about, what is being switched off, what is being tested next, and where budget is moving. Losing creative and losing keywords get cut, winners get more room, negative keyword lists accumulate, audiences are narrowed or widened on evidence, and landing pages get revised where the drop-off is happening on the page rather than in the ad. You get a short written update covering what changed and why — an explanation, not a dashboard link. Not every week produces a dramatic move, and a week where the honest answer is to leave it alone and let it gather data is a legitimate week; saying so is part of the job. The opposite failure is more common and more expensive: accounts fiddled with daily never hold still long enough to learn anything, and the changes end up cancelling each other out while the report fills up with activity.

Once a month the horizon widens. A written report in plain language — what was spent, what came back, what changed, what it means and what happens next — followed by a call to decide the next cycle together rather than to present at you. The report is written against the KPIs agreed in stage three, not against whichever metric happened to look best this month, which is the entire reason for agreeing them in advance. Whatever is not working appears in the report too, along with the recommendation that follows from it, including the uncomfortable ones. Every figure can be checked in your own accounts, because the accounts are yours and your access is never reduced to a screenshot I chose. A report you cannot verify is a story, and stories are what most people are actually buying when they say they have had bad experiences with agencies.