---
title: "The build: the pipeline and the final number | Link & Impulse"
description: "Once the audit has said what needs fixing and the tracking holds: spend, source and outcome joined every night in a database of your own, a dashboard, and the final number, what a customer cost you by channel. Then a retainer that keeps it running with a monthly report."
url: "https://linkandimpulse.com/build"
---

# The build

Once the audit is done and your tracking is fixed, we build your reporting. Every night, your spend, sources and results are brought together in a database you own, with a dashboard on top. It ends with the final number: what a customer cost you, by channel. The retainer keeps it all running.

[Book a call](https://linkandimpulse.com/contact)

- Price

  Priced after [the audit](https://linkandimpulse.com/audit), once we know your systems

- Updates

  Every night, before your office opens

- The number

  Once 60 to 90 days of clean data have built up after [the fix](https://linkandimpulse.com/tracking)

- You keep

  The database, with its full history

- Then

  [The retainer](https://linkandimpulse.com/build#retainer), monthly, priced with the build

## What the build sets up

- ### A database of your own

  Your spend, sources and results, brought together in one place. It holds only your data, and it’s yours to keep, with its full history.

- ### The nightly update

  Every night we pull in data from each of your systems, rebuild every table and check the results before anyone sees them. There’s an example below.

- ### A dashboard

  Cost per customer by channel and by month, plus a page showing the health of your data: every update and every known problem, in plain English.

- ### The final number

  Once 60 to 90 days of clean data have built up after the fix, we write up what a customer cost you, by channel, like the sample below.

Example

## The final number, written up

This is what the build ends with. The retainer sends a report every month, and every third month it includes a fuller quarterly write-up like this one. Below is the sample, section by section, with a note on what each part is for. The business and every figure in it are fictional. Yours follows the same format in your own terms, whether you count clients won, jobs sold or paid treatments. [Download the sample report](https://linkandimpulse.com/sample-audit-report.pdf)

Quarterly write-up

Example Services Group

June to August 2026

Prepared September 4, 2026

Always three _complete_ months. The month you’re in is left out, because half a month looks like a sudden drop.

### Executive summary

Over the three months you spent $104,000 on media and won 44 customers. Directories cost the least per customer and events the most. Client referrals brought in 10 customers with no recorded spend. In August, ad spend rose while customers won fell.

Written from the computed numbers, so it can’t quote a figure your data didn’t produce.

### By the numbers

- Media cost

  $104,000

- Customers won

  44

- Cost per customer

  $3,059

- Tracked calls

  1,412

- Answered

  81%

Cost per customer by channel

| Channel          | Media cost | Won | Per customer |
| ---------------- | ---------- | --- | ------------ |
| Paid search      | $59,300    | 19  | $3,121       |
| Events           | $24,000    | 6   | $4,000       |
| Direct mail      | $13,500    | 5   | $2,700       |
| Directories      | $7,200     | 4   | $1,800       |
| Client referrals | none       | 10  | not measured |

Ad spend and customers won by month

| Month       | Ad spend | Won |
| ----------- | -------- | --- |
| June 2026   | $18,400  | 14  |
| July 2026   | $19,600  | 17  |
| August 2026 | $21,300  | 13  |

Calculated from your data and printed as is. Cost per customer is the period’s total media cost divided by the customers won in it, not an average of the monthly figures. A channel with no recorded spend is left blank, not $0.

Share of site visits, first month against last

| Channel        | June | August | Shift    |
| -------------- | ---- | ------ | -------- |
| Paid search    | 27%  | 31%    | +4.0 pts |
| Organic search | 38%  | 34%    | −4.0 pts |
| Referral       | 9%   | 10%    | +1.0 pts |
| Paid social    | 8%   | 7%     | −1.0 pts |
| Direct         | 18%  | 18%    | 0.0 pts  |

Shows where the mix moved. Not why: the numbers can’t separate the season, the market and the marketing.

### What’s working

- Paid search visits rose 18% in August.
- The answered-call rate rose from 76% to 84% in July.

### What’s broken or risky

- Ad spend rose in August while customers won fell: $21,300 against $19,600, 13 customers against 17.
- Calls after 5 pm and at weekends went unanswered 44% of the time.

Every change, largest first. Rises go in one list, falls in the other, and each one comes with its number.

### Unusual days

- Calls fell 61% on Sunday, July 12.
- Paid search spend rose 48% on Wednesday, August 19.

A day is listed only if it moved 35% or more _and_ was well outside its usual range. Strict on purpose, so every day listed is worth asking about.

### What we’d fix first

Answer the phone before buying more clicks. Route after-hours and weekend calls to a live answering service for 60 days, then compare the answered rate and customers won against this report.

One recommendation, written by a person. The report isn’t finished until it has one.

### Appendix: method and data

Period covered: June to August 2026. Sources: paid search, web analytics, search performance, call tracking, your CRM, and your offline spend sheet. Media cost is paid search spend plus recorded offline spend. Every figure was computed from your own data and reproduced verbatim.

Where every number came from, so nobody has to take our word for it.

[Download the sample report](https://linkandimpulse.com/sample-audit-report.pdf)

PDF, three pages. Printed from the same template the quarterly write-up uses, so it’s easy to pass on to whoever signs off the budget.

Example

## Every night, before your office opens

Each night we pull in data from every system, rebuild every table and check the results before anyone sees them. If one source fails, it’s flagged and the others carry on without it.

Here’s one night, step by step. The steps and their order are real; the times and row counts are illustrative. Each client runs the checks that apply to their systems.

1. 06:30 Run opened for Example Services Group. Its settings are checked before anything else happens.
2. 06:30 Destination confirmed as Example Services Group’s own database. A run pointed at any other stops here, before a row is written.
3. 06:31 Paid search loaded: 1,284 rows.
4. 06:33 Web analytics loaded: 3,910 rows.
5. 06:34 Search Console loaded: 2,206 rows.
6. 06:35 Business Profile loaded: 96 rows.
7. 06:36 Call tracking loaded: 118 calls. Callers’ names and numbers are not kept.
8. 06:38 CRM loaded: 41 new inquiries, 9 deal updates. Contact names are not requested.
9. 06:44 Every table rebuilt from what was loaded.
10. 06:47 Checks run: no failures, 1 warning. Four inquiries typed “Bob’s list” match no channel and are listed on the dashboard.
11. 06:49 Scanned for unusual days: none last night.
12. 06:51 Dashboard refreshed.
13. 06:51 Run closed after 21 minutes. All clear sent.

## The retainer

The retainer keeps the nightly updates and the dashboard running, and adds a report every month, written from the same database. Every third month, that report includes a quarterly write-up like the sample above. It’s priced with the build, and you can stop it at any time.

- ### Cost per customer, by channel

  What you spent on each channel last month, divided by the customers it brought in, in your own word for them: clients won, jobs sold, paid treatments. Events, mailers, directories and referrals get rows of their own.

- ### What your ad platforms counted, compared with your CRM

  The conversions each ad platform counted, next to the inquiries that reached your CRM and the ones that became customers. Where the numbers differ, you’ll see where and by how much.

- ### Missed calls

  Calls nobody answered, by day and hour, and what they were probably worth. There’s an example below.

- ### Tracking health

  The share of new inquiries that arrive with their source attached. If a website change breaks your tracking, this is where we catch it.

- ### One recommendation

  Written by a person. The report isn’t finished until it has one.

Example

## The calls nobody answered

Most offices are staffed nine to five, and people call when they get home. A voicemail at 9 pm often means they ring the next name on the list. Your call tracking already records every call that rang out; the monthly report reads it for you.

Share of calls that went unanswered, by day and hour, over one quarter

Illustrative figures

In office hours, weekdays from 9 am to 5 pm, 10% of calls went unanswered on average. Outside them, 47%. Weekends and evenings after 9 pm were worst.

- Under 15%
- 15 to 30%
- 30 to 45%
- 45 to 60%
- 60% or more
- Office hours

* Unanswered in office hours

  10%

* Unanswered outside them

  47%

* What last month’s 31 missed after-hours calls were probably worth[^c1]

  About $18,600

[^c1]: 31 unanswered calls × a 1-in-5 call-to-customer rate × a $3,000 average customer value. The report uses your own figures.

## Numbers you can trust

Every report we send follows these six rules. [How the numbers are made](https://linkandimpulse.com/how-it-works)

- ### Every figure comes from your own data.

  The written parts of each report are built from the calculated figures, so they can’t include a number your data didn’t produce.

- ### No recorded spend doesn’t mean free.

  A channel with no recorded spend shows a blank, not $0, which would make it look like your best deal.

- ### Recent months are marked as incomplete.

  Some of last month’s inquiries haven’t turned into customers yet, so it isn’t reported as a bad month.

- ### We report what changed, not why.

  The numbers can’t separate the effect of the market, the season and your marketing, so we don’t pretend they can.

- ### Problems are explained in plain English.

  “Four records are in a status we don’t recognise, so they’re missing from the count.” Not an error code.

- ### Estimates are labelled.

  A customer recorded without a value gets an estimate from your usual figures, labelled as one.

## What’s yours and what stays ours

### Yours

The database, with its full history. It holds your data and no one else’s, and if you stop working with us, it’s handed over to you.

### Ours

The software that fills it every night. It stays ours, and it isn’t transferred or licensed to you.

### Stopping

You can stop after any step. If you end the retainer, the nightly updates stop and the database is handed over as it stands.

## Start with a call

We’ll talk through how your inquiries come in and where their source gets lost. If it makes sense, the next step is an audit of your tracking and data, before anything is changed or built.

[Book a call](https://linkandimpulse.com/contact) [How the tracking fix works](https://linkandimpulse.com/tracking)

If an audit wouldn’t be worth it for you, we’ll tell you on the call.