Methodology
How campaigns are built, and how the numbers are counted
Most agency sites assert. This one shows the working, because in a market where anyone can claim anything the only real differentiator left is being checkable. If something below is vague, it is vague because the honest answer is “it depends” and we have said so.
In short
PipelineOS runs Google Ads, Meta Ads, SEO, websites and CRM automation for UK trade and service businesses across 52 sectors. Accounts are created in the client's own ownership. Ad spend goes directly to Google or Meta and is never marked up or charged as a percentage. Measured advertising results quoted on this site come from scaffolding, roofing, glazing and surfacing campaigns only; other sectors carry research and method, and say so.
What we can prove, and what we cannot
This site covers fifty-two sectors. Our own measured advertising results come from four of them: scaffolding, roofing, glazing and surfacing. The screenshots on the results page are from those accounts, uncropped, with the campaign names left visible so the numbers can be placed in context rather than admired in isolation.
Everywhere else on the site, what you are reading is sector research and a campaign method — the job economics, the search intent split, the wasted searches, the seasonality — not a claim that we have run twenty accounts in that field. Those pages are useful because the method transfers and the research is real. They are not a track record, and we are not going to dress them up as one.
If you are in a sector where our own history is thin, the right question to ask on the call is what specifically we would be bringing that a generalist would not. Sometimes the answer is a lot. Sometimes it is that you would be better off with someone who lives in your industry, and we will say so.
The three tiers, plainly
- Measured. Live account data we can open in front of you. Scaffolding, roofing, glazing, surfacing.
- Researched. Sector economics and search behaviour we have worked out and written up, but not yet run an account in.
- Published. Third-party benchmark data, always with the publisher, the year and the sample attached. See benchmarks.
Anything on this site that reads like a result and is not on the results page is one of the last two. We try never to blur them.
What happens before any money is spent
The most valuable part of this is the part that can conclude you should not do it. Four things get established before an account is built, and any one of them can stop the job.
The arithmetic
Average job value, gross margin, and how many quotes you turn into work. Those three numbers produce a gross profit per enquiry, and that number sets the ceiling on what an enquiry is allowed to cost. A firm quoting at £400 a job with a one-in-three close rate cannot survive a £90 lead. A firm quoting at £9,000 barely notices one. Most of the arguments about whether advertising “works” are really arguments about a number nobody calculated.
The demand
Whether the searches exist where you work. Some trades in some towns have real monthly volume; some have eleven searches a month and any budget beyond a few hundred pounds has nowhere to go. This is checked against Google's own keyword data for your actual radius, not a national figure that flatters everybody.
The competition
Who is already bidding, what they are paying, and what their landing pages promise. If four national lead-sellers are bidding your main term, the plan changes: you go around them on the longer, more specific searches rather than paying to lose a head-on auction with somebody who has bought the click to resell it four times.
The leak
What currently happens to an enquiry between the phone ringing and someone quoting. There is no point buying more enquiries into a process that already drops them, and this is the single most common finding. Fixing follow-up is usually cheaper than buying the volume needed to compensate for it.
Want your own four numbers checked?Average job value, close rate, area covered. That is enough to say whether the arithmetic works before anyone builds anything.
Call 07443 392243WhatsAppHow an account actually gets built
| Stage | What is done | Why it is done this way |
|---|---|---|
| Week 0 Tracking |
Conversion tracking installed and tested before a single ad runs: call tracking on the number, form submissions, WhatsApp clicks, and offline import where the quote is what matters rather than the enquiry. | An account without working tracking is not an experiment, it is a donation. Every optimisation decision afterwards depends on this being right. |
| Week 0 Structure |
Campaigns split by the way the money actually differs — high-value work apart from low-value work, commercial apart from domestic, emergency apart from planned — each with matching ad copy and a landing page that answers that intent. | One campaign covering a £400 job and a £9,000 job spends its budget on the cheap half by the middle of the month. The split is the single biggest lever in most trade accounts. |
| Week 0 Negatives |
A sector-specific negative keyword list loaded before launch: DIY and materials searches, training courses, job hunting, jobs the buyer wants for free, and the specific pollution of your trade. | Every sector has searches that can never become a customer. You either know them on day one or you pay to learn them. This is what the sector pages are for. |
| Weeks 1–4 Search terms |
The search terms report read weekly and mined by hand. New negatives added, new ad groups created where a term is earning its own page, bids adjusted by what converted rather than by what got clicked. | This is the actual work. Automation proposes; the report disposes. Accounts that are never read drift towards whatever Google finds easiest to spend. |
| Month 2 Landing pages |
Pages rewritten against what the search terms revealed, not against what we assumed in week zero. Price framing, coverage area and proof placed above the point where people stop reading. | Traffic problems are usually cheaper to fix than conversion problems, so conversion gets fixed second — but it gets fixed with evidence rather than opinion. |
| Month 3+ Compounding |
SEO and Google Business Profile started once paid is profitable, so part of the pipeline stops carrying a click cost. Reviews systematised rather than requested when someone remembers. | Paid search rents attention; organic owns it. Doing organic first when cash flow is the problem is the classic sequencing mistake. |
What counts as a lead, stated precisely
Half the disagreements between a firm and its agency are definitional. So here is the definition, in advance, in writing.
Counted
- A phone call from an ad or the website lasting longer than a set threshold, usually 30 or 60 seconds depending on how your calls run.
- A completed contact form or quote request with real contact details.
- A WhatsApp click that opens a conversation.
- A booked appointment where the business runs on appointments.
Not counted
- Impressions, clicks, page views, or "engagements".
- Calls under the threshold, which are overwhelmingly wrong numbers, existing customers and sales calls.
- Newsletter signups, brochure downloads, or anything else that is a lead only in the sense that a business card is a lead.
The five numbers on every report
- Enquiries as defined above, not conversions in the loose sense.
- Cost per enquiry, calculated on total spend including the wasted part, not on the part that worked.
- Which campaigns and search terms produced them, so budget can follow evidence.
- What was changed since the last report, and why.
- What is not working, named. A report with no bad news in it is a sales document.
Where you can tell us which enquiries became jobs, cost per job replaces cost per enquiry as the headline, because that is the number that actually decides anything.
Where this approach goes wrong
An honest method has failure modes, and pretending otherwise is how you end up with an unhappy client in month four. These are ours.
Thin volume beats good structure. If your trade in your area generates forty relevant searches a month, a beautifully built account still only gets forty chances. No amount of optimisation manufactures demand that is not there, and the right answer is often a wider radius or a different channel.
Tracking can be honest and still be wrong. Call thresholds miscount. People ring the number off the van instead of the website. Somebody finds you on Google, thinks about it for three weeks and types your name in directly, so the credit lands on brand search rather than on the campaign that did the work. Attribution is a best estimate, not a measurement, and anyone presenting it as exact is overselling.
Speed of follow-up outranks everything we do. If enquiries sit for a day, the campaign quality barely matters — the job goes to whoever answered. That is why the CRM and missed-call work usually comes before the ad spend increases, even though it is the less exciting half of the invoice.
Month one looks worse than month three, always. The first month is when you pay to find the negative keywords. A firm that judges the whole exercise on the first four weeks will usually stop just before the point where the numbers turn.
Questions about the method
Do you guarantee a number of leads?
No, and be careful with anyone who does. Search volume in your trade and your area is a fixed quantity that neither of us controls. What we will do before you spend anything is tell you what that volume looks like and what a realistic cost per enquiry would be, so you can decide whether the arithmetic works. A guarantee is either priced with enough padding to make it safe, or it is counting something that is not a real enquiry.
Who owns the accounts?
You do. The Google Ads account, the Meta account, the analytics property, the Google Business Profile and the domain are all created under your ownership with us added as a user. If we part company you remove our access and everything keeps running. Agencies that hold the account in their own name are holding a hostage, and it is the single most common reason firms stay somewhere they have stopped being happy.
How long before it works?
Google Ads produces enquiries in the first week and reliable data in about four. The first month is deliberately noisy because that is when the search terms report is telling you what you did not know. SEO is different: three months before movement is normal and six to twelve before it carries meaningful weight, which is why we do not recommend starting with it if cash flow is the problem.
What if the numbers say no?
Then we say no. Some businesses have a job value too low and a competitive cost per click too high for paid search to work, and some areas simply do not have the search volume. That is a twenty-minute conversation, not a three-month experiment with your money, and we would rather have it at the start.
Related
Also worth reading
- Entity and source maintenanceHow the same discipline is applied to what AI assistants hold about a business.
- Client reviewsWhat clients said, in their own words, with links to the originals.
Talk to a human
Want the arithmetic run on your numbers?
Average job value, close rate and the area you cover. Twenty minutes is usually enough to say whether this is worth doing at all.