# Analytics before ads: the event plan to set up first

> Before spending on ads, track the actions that count as leads, mark them as key events and test them. Without that, no report can tell you which spend worked.

Author: Deb Pradhan
Published: 2026-10-20
URL: https://100xlab.com/blog/analytics-before-ads
Tags: Analytics, Startups

Before you spend on ads, track the actions that count as leads, mark them as key events and test that they fire. Without that, every report shows clicks and none shows which spend brought in a buyer. This post is the event plan we set up first, and the checks that prove it works.

One note before we start: we do not buy media. We build the measurement that every ad budget depends on.

## Why set up analytics before running ads?

Ads buy visits. Analytics tell you which visits turned into leads.

Without tracking, you judge ads by clicks and by gut feel. Clicks are easy to buy. Leads are what you are paying for, and you cannot see them.

There is a second reason. You need a starting point. Every change we make is compared with the numbers before it, and a results check at 30 days only works if the "before" was recorded. If the tracking starts on the same day as the ads, you have nothing to compare against.

So we build the measurement first, because every later decision depends on it.

## What is an event plan?

An event plan is a short list of the actions on your site that you will track. For each action it says what the event is called, when it fires, what details it carries, and whether it counts as a key event.

Here is a shortened version of the event plan for our own site, with the events we mean to flag as key events:

| Event | When it fires | Key event? |
|---|---|---|
| `generate_lead` | The contact form is sent | Yes |
| `book_call_click` | Someone clicks the booking link | Yes |
| `whatsapp_click` | Someone clicks the WhatsApp link | Yes |
| `cta_click` | Any link to the contact page | No |
| `form_start`, `form_abandon` | Someone starts a form, or leaves it unfinished | No |
| `scroll_depth` | A visitor reaches 25, 50, 75, 90 or 100% of a page | No |
| `web_vital` | Page speed readings: LCP (how fast the main content appears), INP (how fast the page reacts to a tap or click) and CLS (how much the layout jumps while loading) | No |
| `js_error`, `page_not_found` | A script error, or a 404 | No |

The key events are the few actions that mean a buyer took a real step. Everything else explains how they got there, or why they did not.

## Which events should count as key events?

Only the actions that mean someone wants to buy or talk: a sent form, a booked call, a started conversation.

Scrolls, hovers and page views are useful context. They are not leads. If you mark them as key events, your reports will look busy and tell you little about leads.

Check the setting itself, too. When we audited our own site in October 2026, our analytics counted zero key events. The events marked as key were ones the site never sends, while the events it did send were not marked. Had we bought ads that week, no report could have shown a single lead from them. The fix is to mark the events that mean a lead, such as `generate_lead`, `book_call_click` and `whatsapp_click`, as key events instead.

## What should you track before your first ad?

Work through this list before any money goes on traffic:

1. **Name the actions that count as a lead.** Often one to three: a form, a booking, a message.
2. **Track each one as an event and mark it as a key event.**
3. **Track the steps before it.** Landing page views, form starts and abandoned forms show where buyers drop off.
4. **Record where each visit came from.** Tag every ad link with UTM parameters (short labels added to the end of a link that name its source and campaign), so they show up in your reports.
5. **Add quality signals.** Page speed, script errors, 404s and rage clicks (three or more fast clicks in one spot), so a broken page is not mistaken for a bad ad.
6. **Keep personal data out.** Our own events never carry names, emails or typed text, and analytics load only after a visitor accepts the cookie notice.
7. **Route every lead to a person.** A lead that sits in a shared inbox for days can go cold, however well the ad worked.
8. **Build one dashboard.** Where buyers come from, where they drop off and what changed this month, in plain words.

## How do you check that the tracking works?

Test it before you trust it. Our checklist:

- **Fire each key event yourself.** Send a test form, click the booking link and click the WhatsApp link. Each should appear as a key event in your analytics' real-time view, and again in the next day's report.
- **Filter out your own visits.** Exclude the team's traffic, so launch-week numbers are buyers, not you.
- **Register custom details.** If your events carry extra details, such as which button was clicked, register them in your analytics so they appear in reports.
- **Repeat after every site change.** A renamed button or a new form can quietly break an event.

## What should the landing page do before you pay for traffic?

Paid traffic lands on a page. If the page is unclear, the ad gets the blame.

The page should answer the buyer's question in its first lines and make the next step obvious. Use the buyer's own words. When people describe their budget, the tool they are replacing and what went wrong last time, those words belong on the landing page and in the ad. We cover writing for both people and AI assistants in [SEO vs GEO](/blog/seo-vs-geo), where GEO stands for generative engine optimisation.

When logic cannot pick the best version, test more than one. For example, ship three: one with the answer up front, one laid out as a comparison table, and one that meets the main objection head on. Keep what the numbers support. Rory Sutherland makes the case in *Alchemy* that cheap tests often beat confident logic, so include the version that looks wrong.

A landing page does not need weeks. One of ours went from blank to live in 3 hours, with analytics running from the moment it went live.

## What happens after 30 days?

Thirty days after go-live, we compare the numbers with the starting point and report what changed. If nothing moved, we say so, and say what we will try next.

That check is the last step of our [method](/method), and it is why the tracking goes in first. Setting up tracking and a dashboard takes about a week, so it fits in one fixed-price Sprint. We compare that with a monthly retainer in [fixed-price marketing vs retainers](/blog/fixed-price-marketing-vs-retainers).

Measurement is the base of marketing engineering, which we define in [what marketing engineering is](/blog/what-is-marketing-engineering). For what we build, see [growth and analytics](/services#growth-analytics).

## FAQ

### Do you run the ads as well?

No. We build the measurement, the pages and the lead routing. Media buying stays with your team or a specialist.

### What is a key event?

A key event is an action you mark in your analytics as one that matters to the business, such as a sent lead form or a booked call. It is the number your ad spend should be judged against.

### How long does it take to set up tracking?

Tracking and a dashboard take about a week as a Sprint. Thirty days later, the results check compares the numbers with the starting point.

### Will we understand the numbers without hiring an analyst?

Yes. You get one dashboard in plain words: where buyers come from, where they drop off and what changed this month.

### We already run ads. Is it too late?

No. Set up the event plan now and test it. Until your key events fire and show up in reports, treat any verdict on the ads with caution.
