The module closes with the practice: how the system is set up in an account, what changes for the sales team and what to expect in the first two months. It is written around Lead Rating (which is always installed alongside Ninja Shield: the complete pack), but it holds for any serious lead scoring implementation.
The pieces of the deployment
| Piece | Where | What it does |
|---|---|---|
| The client's spreadsheet (a copy of the master template) | The client's Drive | Leads, the rating with its breakdown, scoring checkboxes, uploads, configuration, report, manual |
| The spreadsheet's Apps Script (stub + code on the server) | Bound to the spreadsheet | Receives the leads and the actions by webhook (with a token), works out the rating, handles duplicates |
| The visitor snippet | The client's site (WordPress plugin, or a proxy in Next and others) | Fingerprint + GCLID on every form, sent to the webhook |
| The Google Ads script | MCC (installed by the team) | Reads the spreadsheet, uploads Raw/Qualified/Sale in bulk, marks the uploads, feeds back to the Shield |
| The conversion actions | The Google Ads account | All three, created once |
Installation: one day of the team's work, with the client's developer (the snippet) as the only external step — and on WordPress, not even that.
The initial configuration
- The ranking of campaign types (layer C), with overrides by ID where you already know a campaign closes differently.
- Values: qualified (fixed or the rating), sale (an amount from a column or fixed), the raw lead's scaling factor.
- The campaigns' strategy: decide whether the raw lead is primary (only with value bidding) or secondary.
- A written definition of "qualified", shared with the sales team.
- TEST mode: for the first two weeks the rating is worked out and displayed; the uploads can be switched on from day one (the raw lead is harmless as a secondary action) or after validation.
What the sales team sees and does
A tab with the day's leads: name, contact details, campaign, the rating with its breakdown in the note, and three checkboxes. Their job: call the high-rating ones first (the list already comes sorted), tick Qualified when they are, Sale (with the amount) when they close, Bad when they are worthless. Nothing else. That habit is what holds up the recalibration and the offline conversions; without it, the system stays a prediction.
The first 60 days
| Week | What happens | What to look at |
|---|---|---|
| 1-2 | Leads being scored; raw leads uploading with a value | Is the GCLID arriving? Do the ratings make sense (the breakdown)? Is sales marking? |
| 3-4 | First qualified leads and sales uploaded; Smart Bidding in learning if the raw lead is primary | Primary conversions fall (normal); the first points on the calibration curve |
| 5-6 | Budget shifting towards high-rating sources | Cost per qualified lead (it should fall); Google's CPA (it may rise) |
| 7-8 | The first complete monthly report | Calibration curve; the suggested reordering of the ranking; lift per signal (if there is a sample) |
At 60 days: the decision on moving the campaigns to value bidding if they were not already, and adjusting the source ranking with the data.
Symptoms that something is not right
- Lots of leads with no GCLID: the snippet is not capturing it (redirects, third-party forms) or auto-tagging is switched off.
- All the ratings looking alike: the signals are not arriving (the snippet is badly installed) or the weights are flat.
- A flat calibration curve after hundreds of leads: review the signals and the weights; perhaps the definition of "qualified" has changed.
- Nobody ticking the checkboxes: the problem is not technical; with no truth there is no system.
- Uploaded conversions rejected: dates earlier than the click, outside the window, invalid GCLIDs.
What the system does not do
- It does not call the leads and it does not decide what a customer is: it scores, and it learns from what sales mark.
- It does not take the leads out of the client's spreadsheet.
- It does not replace basic measurement: if the form fires no conversion and stores no GCLID, there is nothing to score.
💡 Ninja trick: Lead Rating and Ninja Shield are always installed together because they share the snippet, the fingerprint and the feedback circuit: the Shield cleans up the traffic coming in, the Rating scores the leads that traffic produces, and the good leads improve the score of the sites they came from. Two products, one system — and the declared goal of the 2027 platform: a new client operational in 15 minutes, not in a day.
What you should remember
- The pieces: spreadsheet + Apps Script with a webhook + snippet + Ads script + three actions.
- Initial configuration: source ranking, values, strategy, the definition of qualified, TEST mode.
- The sales team: call by rating and tick Qualified / Sale / Bad.
- 60 days: GCLID → curve → budget shifted → report; clear symptoms if something goes wrong.
That is the end of Module 6. Module 7 goes back to the catalogue: advanced feeds and automation for shops with thousands of products and several markets.