End-to-end analytics: seeing where the money comes from
Ordinary analytics shows clicks and requests. End-to-end analytics shows money. The difference is that the first can present a channel with a hundred cheap requests as your best one, even though not a single one of them turned into a payment.
Ordinary analytics shows requests; end-to-end analytics shows money, by linking an ad click to a specific deal and its value. Without that link you optimise the number of requests rather than profit, and those often pull in opposite directions. You can start without expensive platforms: tags in links, a form that passes the source, and a spreadsheet.
In this article8 sections
The problem it solves
A typical picture: advertising produces requests at $10, maps at $25. The conclusion suggests itself — put more into advertising. But look further down the chain and it may turn out that one in twenty advertising requests buys, while one in three from maps does. In that case a client from advertising costs $200 and one from maps costs $75.
End-to-end analytics links an ad click to a specific deal and its value. Only after that do budget decisions rest on something solid.
Without a link to sales you optimise the number of requests, not profit. Those are different goals, and they often point in opposite directions.
What it consists of
- A source tagEvery visit carries information with it: where the person came from, on which query, from which ad.
- Capturing the enquiryThe request, call or message lands in your system together with that tag.
- Recording the dealThe CRM notes how the enquiry ended and for what amount.
- Bringing it togetherSpending on a channel is matched against revenue from clients who came through it.
The weak link is usually the third: requests get recorded, outcomes do not. The chain breaks halfway and the whole structure stops working.
What you need technically
| Element | What for | Required? |
|---|---|---|
| Analytics on the site | To see behaviour and sources | Required |
| Tags in ad links | To tell campaigns and ads apart | Required |
| A form that passes the source | To link the request to a channel | Required |
| A CRM or a spreadsheet | To store the deal outcome | Required |
| Dynamic phone numbers | To identify the source of a call | Desirable |
| A ready-made platform | To automate the reconciliation | As you grow |
How to start without expensive systems
Full end-to-end analytics platforms cost money and take setting up. You can start far more simply and almost free — and that is enough for most small companies.
- Put tags in every advertising link.
- Configure the form so the source is passed along with the request.
- Start a spreadsheet: date, source, contact, status, amount.
- Once a month, reconcile channel spending against revenue from that sheet.
- When enquiries pass a few dozen a month, move to a CRM.
A spreadsheet filled in with discipline is more useful than an expensive system nobody enters deal outcomes into. The problem is almost never the tool — it is the habit.
What to do about calls
Calls are the hardest channel to track: the person phones and the link to the source is gone. There are solutions, and they differ in price.
- Ask during the conversation. Free but imprecise: people do not remember.
- Different numbers for different channels. One for the site, one for maps, one for business cards. Simple and cheap.
- Dynamic numbers. Each visitor is shown their own number and the source is identified automatically. Precise, but paid.
What to look at in the end
Not twenty charts, but a handful of numbers per channel.
| Metric | What it tells you |
|---|---|
| Spend | How much went in |
| Requests | How many enquiries you got |
| Cost per request | How expensive an enquiry is |
| Clients | How many reached payment |
| Revenue | How much money the channel brought |
| Return on spend | Revenue relative to cost |
This table, filled in once a month per channel, is enough to manage a budget. Everything else is detail.
The multiple-touch problem
A client rarely arrives in one go. They saw the ad, found you in search a day later, read reviews on maps the day after that, and then called. Which channel gets credit for the sale?
Simple systems credit the last one — and that distorts the picture: the advertising that brought the person in first looks useless, while maps take all the credit.
| Model | Who gets credit | When it applies |
|---|---|---|
| Last touch | The channel before the purchase | Short cycle, simple purchases |
| First touch | The channel that brought them in | Judging acquisition channels |
| Linear | Split evenly across everyone involved | Long cycle, many touches |
For a small business it is enough to look at both extremes: first touch and last touch. If a channel looks bad in one and good in the other, it works at the top of the funnel and must not be switched off.
What to start with tomorrow
- Check whether tags are applied to every advertising link.
- Make sure the form passes the source together with the request.
- Start a spreadsheet and enter the last month's enquiries into it.
- Mark the outcome of each one — from memory if need be.
- Calculate cost per client by channel. The picture almost always differs from what you expected.
Frequent questions
Do we need an expensive platform?
No. Up to a few dozen requests a month, tags, a form and a spreadsheet are enough. A platform pays for itself when manual reconciliation starts eating serious time.
What if sales happen offline?
Record the source at the point of enquiry and enter the outcome by hand. It is less precise but gives you the main picture. Dynamic numbers improve the accuracy.
How long does setup take?
The basic version, a few hours. A full system with a CRM and dynamic numbers, from a few days to a couple of weeks.
How do we handle a long sales cycle?
Count from the date of first enquiry, not the date of payment, and read the data with a lag equal to your cycle. Otherwise recent months will always look worse than they are.
What if the numbers from different systems do not match?
A 10–30% discrepancy is normal: blockers, cookie refusals, different counting models. It is worrying when the difference is a multiple — then look for a setup error.
Do we need analytics if we have few requests?
Especially then. At ten requests a month every one counts, and knowing where they come from matters more than it does at a hundred.
The materials answer general questions. We will look at your specific case — free and without obligation.