SEO and paid advertising

Google Ads Support

An account spending steadily with nobody looking at search terms, so budget goes to clicks that were never going to convert.

What it looks like in practice

The account has been spending steadily for months and the phone has not got busier. Open the search terms report and the answer is sitting there: the budget went to people looking for jobs, free templates, and one competitor's warranty policy.

How I build it

1. Check conversion tracking before touching anything else, because the rest is guesswork without it.
2. Read the search terms report and add negatives for what should never have matched.
3. Group keywords tightly enough that the ad can speak to the search.
4. Write ad variations that differ in claim, not in wording.
5. Report weekly on leads and cost per lead, not clicks.

What is included

• Keyword and ad group notes
• Ad copy variations
• Landing page notes
• A performance summary that names next actions

The first hour on any account

1. Does a conversion fire when the form is submitted? Test it.
2. Search terms for the last 30 days, sorted by spend. Add negatives.
3. Are calls counted as leads only when they last long enough to be real?
4. Does the ad match the page it lands on?
5. What is the cost per qualified lead, not per click?

Where it usually goes wrong

• Broad match left running with no negatives
• Conversions counted as button clicks rather than real enquiries
• Ad groups wide enough that the ad cannot speak to the search
• Performance judged on cost per click, which nobody has ever banked

What it is like once it works

Spend goes to searches from people who could plausibly buy, the ad matches what they typed, and the weekly number being watched is cost per qualified lead rather than clicks.

You probably need this if: Clicks are up and the phone is not ringing; Nobody has opened the search terms report.

SEO Content Optimisation

A page written for the business rather than the reader, missing the words people actually type.

What it looks like in practice

The page is well written and it never appears in a search. It says solutions where customers say repairs, and it says we where the reader wants you. The headings are decorative, so a search engine and a person skimming get the same nothing out of them.

How I build it

1. Find how the audience phrases the problem, not how the industry names it.
2. Give the page one subject and one search intent.
3. Fix the structure: one h1, headings that describe sections, short paragraphs.
4. Write the title and description to be clicked, inside the length limits.
5. Link to and from the pages that belong with it.

What is included

• Title options
• Meta description
• Heading structure
• Internal link suggestions

Length limits worth respecting

• Title: around 60 characters, or search results cut it off
• Description: around 155 characters
• One h1 per page, then h2 and h3 in order with nothing skipped

These are not ranking tricks. They control what a person sees before they decide whether to click.

Where it usually goes wrong

• The page targets a phrase nobody outside the industry uses
• Headings are styled for looks rather than written to describe
• The same keyword is pushed into a page that was never about it
• The title tag is the business name on every single page

What it is like once it works

The page uses the words the customer uses, the headings describe what follows, and the first paragraph answers the question rather than introducing the company.

You probably need this if: Pages rank for your brand name and nothing else; Your headings are decorative rather than descriptive.

Keyword Mapping & Topic Plan

Content written on instinct, where two pages compete for the same search and the topics that matter have nothing at all.

What it looks like in practice

There is a blog. Nobody can say why any given post exists. Two of them compete for the same phrase, the service that actually pays the bills has no page at all, and the next post will be chosen the way the last twelve were, by whoever had an idea on the day.

How I build it

1. Collect the questions the sales calls keep answering.
2. Group them by intent: ready to buy, comparing, still learning.
3. Assign one group to one page, so nothing competes with itself.
4. Note what already exists, what needs rewriting and what is missing.
5. Order the list by commercial value, not by search volume.

What is included

• Keyword groups
• Search intent notes
• Page recommendations
• A priority order

One row of the map

Intent decides the page type. Sending a comparison search to a sales page is why so much content never converts.

Where it usually goes wrong

• Topics picked by search volume rather than commercial value
• Two pages targeting one phrase, splitting their own results
• No note of what already exists, so work is repeated
• A plan with no owner and no order, which becomes a list

What it is like once it works

One page owns one intent, the plan is ordered by what the business needs to sell, and writing the next piece begins with a row that already exists rather than a blank page.

You probably need this if: You have a blog with no plan behind it; Two of your pages rank for the same phrase.

Local SEO Page Support

A business that serves an area but has one page for everywhere, so it competes with nobody in particular.

What it looks like in practice

You cover five suburbs and rank in none of them. The site has one page that lists all five in a sentence, which is precisely what every competitor within twenty kilometres has also done.

How I build it

1. Confirm the service areas that are worth a page of their own.
2. Write genuinely local content, not a template with the city swapped.
3. Keep name, address and phone identical everywhere they appear.
4. Support the pages with the Google Business Profile and real citations.
5. Add the questions local customers actually ask.

What is included

• A city and service page outline
• Local keyword placement
• FAQ ideas
• A call to action structure

What makes a location page real

Landmarks, suburbs and travel times you would only know locally. Work you have done nearby. The regulations or conditions specific to that area. A photograph taken there.

If the page reads correctly after swapping the city name for another, it is a template, and it will perform like one.

Where it usually goes wrong

• Location pages that are one template with the suburb name swapped
• No mention of anything that proves the business has been there
• The map listing and the website disagree about the phone number
• Pages created for areas the business will not actually travel to

What it is like once it works

Each area worth winning has a page with something real on it, the listing and the site agree, and the business turns up in the searches that happen a few streets away.

You probably need this if: You serve five suburbs and rank in none; Your map listing is missing from searches you should win.

Meta Ads Creative Testing Plan

Ads that change one word at a time, so nothing conclusive is ever learned and fatigue sets in before the answer does.

What it looks like in practice

Ads run well for a fortnight and then stop. The response is to change the headline and try again, which is why after six months nobody can name a single thing that has been learned about what this audience responds to.

How I build it

1. Write down the belief each creative is testing, before making any of it.
2. Vary the angle rather than the adjective: different problem, different proof, different format.
3. Keep audience and placement constant while creative is the variable.
4. Give each enough budget and time to say something.
5. Keep the winner, replace the losers, and record what was learned.

What is included

• Hook variations
• Primary text options
• A creative angle list
• Call to action recommendations

Four angles worth testing first

1. Problem. Name the pain in the first three seconds.
2. Proof. Lead with a result or a client quote.
3. Process. Show how the work is done, which builds trust for considered purchases.
4. Objection. Answer the thing that stops people buying.

Four angles teach you more in two weeks than forty variations of the same headline.

Where it usually goes wrong

• Changing several things at once, so the result explains nothing
• Stopping a test before it has enough data to mean anything
• Testing wording when the angle is what is wrong
• No place where results are written down, so they are relearned

What it is like once it works

Each test asks one question, the answer is written down, and the next round starts from what the last one proved rather than from another idea.

You probably need this if: Your ads stop working after a fortnight; Testing means changing the headline.

Conversion Tracking QA Checklist

Ads running against tracking nobody has verified, which means the reporting is fiction and optimisation is worse than guessing.

What it looks like in practice

The platform reports forty leads. The inbox contains eleven. Somebody has been optimising toward the forty for three months, which means the budget has been steered by a number that was never real.

How I build it

1. List the actions that count as a conversion for this business.
2. Test each one by performing it, not by reading the configuration.
3. Check for duplicates, where a single action fires two events.
4. Confirm phone calls count only when they were long enough to be real.
5. Re-test after any website change, because this breaks silently.

What is included

• An event checklist
• Form submit verification
• Button click checks
• Conversion goal recommendations

The checklist I work through

• Form submit fires once, and only on success
• Thank-you page or state is reached
• Phone click tracked on mobile
• Call conversions have a sensible minimum duration
• WhatsApp or chat clicks tracked if they are a real route
• No test traffic counted, including your own
• Ad platform and analytics broadly agree

Where it usually goes wrong

• Nobody has submitted the form themselves since the site changed
• Page views counted as conversions, so everything looks successful
• The same lead counted twice by two different tags
• Phone calls counted the moment they connect, including wrong numbers

What it is like once it works

Every conversion has been fired by hand at least once and traced to where it lands, so the number in the platform matches something you can point at.

You probably need this if: Your ad platform reports more leads than your inbox shows; Nobody has tested tracking since the site changed.

Campaign Performance Report

A monthly export nobody reads, full of numbers with no interpretation and no decision attached.

What it looks like in practice

The report goes out monthly: forty pages of exported charts, opened about as often as a set of terms and conditions. Nobody has ever made a decision because of it, and nobody wants to say so.

How I build it

1. Report the same figures every period so the trend is readable.
2. Lead with the number that matters to the business, usually cost per qualified lead.
3. Explain what changed, in one paragraph, in plain words.
4. Name what you will do next, and what you need from the client.
5. Keep the detail in an appendix for whoever wants it.

What is included

• A metric summary
• Top performing assets
• Issue notes
• Next step recommendations

The shape of a report a client actually reads

1. Headline. Leads, cost per lead, against last period.
2. What moved. Two or three sentences.
3. What worked. The assets worth keeping.
4. What I am changing. Specific, with a date.
5. What I need. Access, approval or content.

A report that names the next action gets read. One that only lists metrics gets filed.

Where it usually goes wrong

• Charts presented with no sentence saying what they mean
• The definition of a metric changing quietly between months
• Good news at the top and the useful part on page nine
• No next action, so the report ends in agreement rather than a decision

What it is like once it works

One page: what happened, what it means, what is being done next. The detail still exists underneath for anyone who wants it, and the covering note is short enough to read in a lift.

You probably need this if: Your reports are dashboards nobody opens; You cannot say what changed after last month's report.