๐ AI Lead Scoring ยท How to Filter Noisy Leads
This article continues Detail 1๏ธโฃ ยท How to Search Massive Leads ยท AI Database. After a search lands a pile of leads, don't save them all immediately โ follow the steps below to filter out the noise. Otherwise you'll waste points and email resources.
๐ Core goal: Remove noisy leads, so you don't waste points on downstream emails.
No matter the search approach, never select all and save straight off the results. Preview and filter first to guarantee precision and protect your follow-up reply rate.
How Should I Filter?โ
โถ๏ธ Watch the video below, or follow along with the doc โ you'll master every filtering approach in a few minutes:
๐บ Video blank? Open in a new window โ
I. Step 1 ยท Skim for Precisionโ
After landing on the results page, don't start from page 1 and grind through โ exhausting, and it tells you nothing about overall precision.
The right move: Sample pages 1~5 and the last few pages. In 30 seconds you can tell whether the batch is worth saving.
Precision benchmarks:
| Sample result | Action |
|---|---|
| Front/back pages โฅ 80% precise | โ No filtering โ save directly (recommend โฅ 70% before saving) |
| Front/back pages < 80% noisy | ๐ Run the detailed filtering in steps "II" and "III" |
II. Step 2 ยท Pick a Country Scopeโ
Before any filtering approach, decide: Do I want to differentiate by country?
1. Country-Agnostic (Standard Approach)โ
You can skip country segmentation and run a unified English outreach campaign for marketing. All you need to do is exclude domestic competitors.
๐ก Recommendation: create a baseline view that excludes China.
Stepsโ
- โ Click Filter at the top of the results list โ Country or Region
- โก Pick Does not include
- โข Check the countries/regions to exclude (China, India, Pakistan, etc.)
- โฃ Click Apply Filter
- โค Save as a named view
2. Country-Specific (Detailed Approach)โ
If you'll run country-specific marketing later โ localized outreach emails (German, Portuguese, etc.) โ check your target countries, save as a view, and manage them downstream.
Stepsโ
- โ Click Filter โ Country or Region
- โก Pick Any of
- โข Check the target countries/regions (the search box helps you locate fast)
- โฃ Click Apply Filter
- โค Save as a named view
Step โข: Search for the country/region you want to develop
Steps โฃ + โค: Click Apply Filter to lock in the conditions, then save as a view
๐บ Walkthrough video: How to segment by countryโ
๐บ Video blank? Open in a new window โ
III. Step 3 ยท Pick a Filtering Approachโ
With the country scope locked in, filter further within that view. The system offers 2 base approaches + 1 combined approach. Use the decision tree above to pick what suits you:
1. ๐ Manual Filtering (plenty of time ยท want to save points)โ
Stepsโ
- โ Preview results; for obviously mismatched customers, click Blacklist at the end of the row
- โก Identify recurring "noise keywords" (e.g., "retailer", "second-hand")
- โข Add the noise keywords to the view's Does not include condition
- โฃ Click Save and Preview
Step โ : For mismatched customers, click Blacklist at the end of the row
Steps โก + โข + โฃ: In the target view, click Filter in the top right โ pick Chinese Description โ Does not include โ type "retailer" and press Enter โ click Save and Preview
Below the filter conditions, pick "Meets all of the above". Otherwise hitting any one condition leaves the row in, defeating the exclusion.
2. ๐ค AI Lead Scoring (limited time ยท plenty of points)โ
Before using AI Lead Scoring, set up your product profile. The system uses the profile to judge whether each company is a potential customer and gives a match score (0~100). Higher = better match.
Use it like thisโ
| Your situation | Recommended action |
|---|---|
| First few pages look clearly precise | Skip scoring; sample and save directly |
| Current page precision is uncertain | Randomly pick 30~50 customers for AI Lead Scoring (cost: 30~50 points) |
| Most scores match | Safe to save this slice of data |
| Most scores don't match | Switch customer group, change keywords, or narrow the range |
Stepsโ
- โ In Settings - Product Profile, fill in your product (not the customer's product)
- โก In the results list, check the customers to score
- โข Click the AI Lead Scoring button in the top right of the results list
- โฃ Fill in the scoring rules in the popup; create the task
- โค Open Batch Scoring Tasks to view results
Steps โก + โข: Check customers, click AI Lead Scoring
Step โ : Product profile reference (fill in your product, not the customer's)
Step โค: Open Batch Scoring Tasks (a record of all AI Lead Scoring history)
Click View detailed report at the end of a row to see AI's deep analysis for each customer:
This section is just an intro for new users. 4 entry points (AI Database batch / single insight / list hover / LinkedIn) + 6 scoring dimensions + 4 ways to save points โ see ๐ ๐ค AI Lead Scoring Guide
๐บ Walkthrough video: Manual Filtering vs AI Lead Scoringโ
๐บ Video blank? Open in a new window โ
3. ๐ฅ Combined Approach (middle ground)โ
Manual filter the first slice + AI-score the rest โ balances points and time. Most experienced users go this way.
Short version: preview the early results, manually pick and save the most precise, then AI-score the noisier rest.
๐ Full steps in Scenario 3: Manual + AI combined save
๐ Common Pitfalls in the Filtering Phaseโ
| Pitfall | Consequence | How to avoid |
|---|---|---|
| ๐ด Select all โ AI Lead Scoring straight off the bat | Burns hundreds to thousands of points instantly; can't be cancelled | Sample first to judge precision, then decide which slice gets AI-scored |
| ๐ก Saving at 70% precision | Wastes points / hurts domain reputation downstream | Recommend โฅ 80% precision before saving |
| ๐ก Flipping pages too fast / clicking constantly | System fails to load | Give it a few seconds per page |
| ๐ก Forgot to pick "Meets all of the above" in filters | Exclusion keyword doesn't take effect | Check the logic toggle at the bottom of the filter popup |
| ๐ก No product profile before AI Lead Scoring | Inaccurate scoring | Fill in Settings - Product Profile first |
Once you've filtered down to qualified data, move on to saving. What should you watch out for when saving?
๐ Detail 3๏ธโฃ ยท How to Save Qualified Leads ยท Contact Saving โ Tag naming formula + 3 saving strategies
๐ Permalink: https://laifa.xin/zhinan/02-filter-customers