Choosing a traffic filtering service by the promises on its homepage is pointless: everyone writes “best protection” and “zero bots”. The real differences are in the details: how the service explains its decisions, how easy it is to catch its mistakes, how it connects to your site and what it can do after filtering. Below are the criteria that actually set one bot protection service apart from another, with questions to test each one.
What a traffic filtering service actually does
On every visit from an ad link, the service collects signals in a fraction of a second (network and ISP, country, device, browser traits, behavior) and makes a call: the offer for a real person, or a neutral page for a bot, an ad review crawler, a spy tool or an untargeted visitor. The mechanics are covered in what is cloaking, and the filtering methods in how to filter out bots. How such a service works under the hood is explained in cloaking service explained.
A service can get it wrong in two directions:
| Mistake | What happens | What it costs you |
|---|---|---|
| Let a bot through | Automated traffic reached the offer | Polluted stats, questions from the network, exposed creatives |
| Blocked a human | A buyer saw the neutral page | A lost conversion on a click you already paid for |
The second mistake is the nastier one: you do not see it until you go looking for it. So the main selection criterion is not “how many bots it catches” but “how easy it is to see and fix a mistake”.
Criterion 1. Transparency: a reason for every click
A service that only says “passed” or “blocked” cannot be tuned. You see that half your traffic was filtered, but not why, and you cannot tell honest filtering from a broken rule.
What you should get:
- A decision reason for every visit: a clear label like “VPN or proxy”, “Country not on the allowlist”, “Per-IP click limit exceeded”, not a three-letter code.
- A visit card with the network, ISP, device, ad click parameters and which checks fired.
- A breakdown of reasons over a period: if almost all filtering comes from one reason, that is the one to check.
- A full breakdown of a single visit, so you can show support a concrete example of a disputed decision.
How to read these reasons in practice is covered in why a click went to the White Page.
Criterion 2. Accuracy without cutting real users
A strict filter catches more bots, but also more people. A good service gives you tools to find the balance for a specific source, not a single “maximum protection” button.
What to look for:
- Per-platform settings. The Facebook in-app browser drops the referrer, TikTok has its own ISP patterns, native ads produce a lot of repeat impressions. The service should account for this itself or at least let you configure it.
- Strictness levels with a clear description of what each one adds.
- Trust scoring, not just hard blocks. One suspicious signal should not block a visit if everything else looks fine. This matters most for mobile networks and in-app browsers.
- Measuring the loss before a rule goes live. A mode where the filter only marks who it would have blocked but lets everyone through, or a replay of past traffic against new settings.
- An honest take on pass rate. The share of visits that reach the offer is not a grade on its own: paid traffic with a verified click ID can have a high pass rate, and that is normal. The warning sign is when almost nobody reaches the offer, because that means the rules are cutting real people. A service that sells a low pass rate as “great protection” is misleading you.
Tip. Click your ad link a few times from your own phone on mobile data and find those visits in the log. If your own visit was blocked without a clear reason, the same thing will happen to buyers.
Criterion 3. Decision speed
While the service thinks, the visitor waits. On mobile data every extra fraction of a second means some people close the tab before the page appears.
What to ask:
- How long a decision takes, and whether that time is published anywhere besides marketing copy, for example on a system status page.
- What happens if the service does not respond: where the visitor goes and how long they wait.
- Whether there are extra checks that add waiting time, and whether you can enable them only where they are needed.
Keep the trade-off in mind: the deepest browser checks take time. A good service lets you choose: fast network checks for lenient sources, and an extra behavioral check for strict ones.
Criterion 4. Setup methods: your site stays with you
The setup method affects both convenience and risk. The sensible model is one where your landing page stays on your server and your domain, and the service only makes the decision on each visit. How this differs from a PHP cloaking script on your own server is covered separately.
Common options:
| Method | How it works | When it fits |
|---|---|---|
| JS tag | A snippet as the first line in the landing page <head> |
Any site, including website builders |
| PHP file | A file on your server asks for a decision before the page is shown | PHP hosting, when you want full control |
| CMS plugin | The same server-side method, installed from the admin panel | WordPress sites |
| Inside your tracker | The filter runs inside your tracker | Traffic already goes through Keitaro or Binom |
Both main methods are covered step by step in how to connect a cloaker to your site.
Questions for the service: can you verify the connection from the dashboard, what happens if the code fails to load, is the server file updated, and how will you learn about a new version?
Criterion 5. Built-in tracking
Filtering answers “who came”, tracking answers “how much did it earn”. If the service only does the first, you will have to run a separate tracker and wire the two together.
What is useful to have in one place:
- receiving postbacks from affiliate networks and attributing conversions to clicks;
- cost: from the link, by pricing model or entered manually, plus profit and ROI;
- campaign parameters without a hard limit, so you can tag creative, audience, placement and buyer;
- sending conversions back to ad platforms;
- integration with external trackers if you already work in one.
Why you need a tracker at all and how it differs from a filter is covered in tracker for affiliate marketing and cloaker vs tracker: what is the difference.
Criterion 6. Reports you can act on
A pretty dashboard is worthless if it does not tell you what to switch off. Useful reports answer specific questions:
- which placements, creatives and geos make money and which only burn budget;
- where traffic was filtered: at the network checks, at the browser check, or because the check never came back;
- how pass rate and conversion changed day by day;
- whether you can export the data and build your own breakdown.
Check the history depth too: how many days back you can view clicks and how that depends on the plan.
Criterion 7. Team and access
If you do not work alone, you need:
- roles with different permissions: someone sets up flows, someone only views statistics, someone sees finances;
- projects, so a buyer only sees their own campaigns;
- guest links to a report for a client without dashboard access;
- an audit log and login protection: two-factor authentication and session management.
How to organize team work is covered in media buying as a team.
Criterion 8. Support
A filtering problem is almost always urgent: the campaign is running and money is going out. Find out in advance:
- which channels you can use: tickets, messenger, email;
- whether you can attach a specific click or visit breakdown to a request;
- whether there is documentation describing every setting and the order of checks;
- whether there is a system status page showing if the service is working right now.
A service that hides how it decides behind a “secret algorithm” always wins the argument. A service that shows the reason and the order of checks lets you find the mistake yourself, whether it is yours or theirs.
Red flags when choosing
Some promises should put you on guard rather than win you over:
- “Ban protection guaranteed”. An ad account ban depends not only on filtering but also on creatives, the landing page, domain, offer and account history. An honest service lowers the risk but cannot guarantee an outcome on behalf of the platform.
- “We block 90% of traffic, so the protection works”. A high block rate is not a virtue by itself. What matters is who exactly was blocked.
- No decision log. If you cannot look at an individual visit, you cannot verify a single claim the service makes.
- A requirement to move your site or domain. Your pages should stay with you: that way you do not depend on someone else's infrastructure and do not share domain reputation with other people's projects.
- No documentation. Without a description of settings and the order of checks, every change becomes an experiment.
How ArtisanClo handles it: what we show openly
We built ArtisanClo around these same criteria.
- A reason for every click. In the click log every visit has one of dozens of clear reasons and a card with the network, device and checks. Statistics shows at which of three steps traffic was filtered: network and request, browser check, or check never came back.
- The order of checks is published in the documentation: which condition fires before which, and what the visitor will see.
- Platform-specific protection. The traffic source tunes the checks for Facebook, Google, TikTok, native ads and other platforms, and on top of that you choose a strictness level: Soft, Balanced or Strict.
- Measuring losses. Shadow mode, available with the PHP file setup, lets everyone through and marks who the filter would have blocked. Diagnostics replays past visits against the current settings and shows which filter is cutting traffic and what it costs.
- Setup without moving your site. A JS tag or a PHP file on your server, or the WordPress plugin; for Keitaro and Binom, a filter and gateway inside the tracker. Your site and domains stay with you.
- Tracker included. Postbacks from networks, cost, profit and ROI, unlimited SubID parameters, sending conversions to ad platforms.
- Team. Roles with per-section permissions, projects, guest report links.
- Support and status. Tickets and Telegram, documentation, and a public system status page that shows the decision time.
The full list is on the features page. There is a free trial; terms and plans are on the pricing page.
Checklist before you pay
- Can you see the decision reason for every click?
- Can you see who the filter would block without losing traffic?
- Are there settings for your traffic source?
- Did you find your own test click from a mobile phone in the log?
- Does the service connect without moving your site and domain?
- How long does a decision take, and where can you see it?
- Does the service accept postbacks and calculate ROI, or do you need a separate tracker?
- Can you break down revenue by placement, creative and geo?
- Are there roles and projects for a team?
- How fast and through which channels does support respond?
The bottom line
You choose a traffic filtering service not by how loud its promises are but by how transparent it is: whether the reason behind every decision is visible, whether you can measure losses before a rule goes live, how it connects and what it does after filtering. Test it on a small real campaign, find your own clicks and the disputed decisions in the log, and within a few days it will be clear whether it fits your traffic. Common mistakes that do not depend on which service you pick are collected in cloaker not working: how to check and fix it, and a short list of pre-purchase questions is in buying a cloaker wisely.
