Analytics and money

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.

12 min read Updated 30 August 2026
In short

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
  1. The problem it solves
  2. What it consists of
  3. What you need technically
  4. How to start without expensive systems
  5. What to do about calls
  6. What to look at in the end
  7. The multiple-touch problem
  8. What to start with tomorrow

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

  1. A source tagEvery visit carries information with it: where the person came from, on which query, from which ad.
  2. Capturing the enquiryThe request, call or message lands in your system together with that tag.
  3. Recording the dealThe CRM notes how the enquiry ended and for what amount.
  4. 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

ElementWhat forRequired?
Analytics on the siteTo see behaviour and sourcesRequired
Tags in ad linksTo tell campaigns and ads apartRequired
A form that passes the sourceTo link the request to a channelRequired
A CRM or a spreadsheetTo store the deal outcomeRequired
Dynamic phone numbersTo identify the source of a callDesirable
A ready-made platformTo automate the reconciliationAs 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.

  1. Put tags in every advertising link.
  2. Configure the form so the source is passed along with the request.
  3. Start a spreadsheet: date, source, contact, status, amount.
  4. Once a month, reconcile channel spending against revenue from that sheet.
  5. 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.

MetricWhat it tells you
SpendHow much went in
RequestsHow many enquiries you got
Cost per requestHow expensive an enquiry is
ClientsHow many reached payment
RevenueHow much money the channel brought
Return on spendRevenue 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.

ModelWho gets creditWhen it applies
Last touchThe channel before the purchaseShort cycle, simple purchases
First touchThe channel that brought them inJudging acquisition channels
LinearSplit evenly across everyone involvedLong 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

  1. Check whether tags are applied to every advertising link.
  2. Make sure the form passes the source together with the request.
  3. Start a spreadsheet and enter the last month's enquiries into it.
  4. Mark the outcome of each one — from memory if need be.
  5. 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.

Still have questions about your business?

The materials answer general questions. We will look at your specific case — free and without obligation.