What is Direct traffic?
Direct is a term traditionally used in marketing and is synonymous with “unknown origin.”
Which links are usually Direct?
Generally speaking a link encountered in a web browser, when clicked, will broadcast the site on which the link is hosted and the target site (your site) can pick that up. All other links are classified as Direct unless it is marked up with query parameters.
Links which are not likely to be classified Direct
Links found on websites in a mobile or desktop web browser (with certain exceptions, below)
Links which you have marked up with utm or other query string parameters which are matched by rules in your Channel Assignments
Links which are likely to be classified as Direct
Links on websites which manually remove referrer information
Most email sites (Gmail, Outlook, iCloud Mail)
Most messaging sites (Discord, WhatsApp Web)
Other sites with tight security policies, like those handling personal informartion
Links on any site, in any browser with a security setting to remove referrer headers
Links within mobile apps
Search apps (Google, Bing)
Social apps (Facebook, Instagram, Youtube)
Email apps (Gmail, Mail by Apple, Outlook)
Messaging apps (SMS, Messages, Facebook Messenger, WhatsApp, Teams, Slack)
AI assistants (Siri, Google Assistant, Alexa, Gemini, ChatGPT)
Links within desktop applications
Email clients (Outlook, Apple Mail)
Documents (Microsoft office documents, PDF)
Desktop Messaging clients (Teams, Slack, Facebook Messenger, WhatsApp, Messages by Apple)
Smart home assistants
Browser bookmarks
When a user has visited your site before and they start to type your website address into a browser address bar using autofill
If Corvidae stitches together longer, more accurate user Journeys, why do I still see a lot of Direct?
In traditional attribution the length and coverage of a customer journey is limited to the fidelity of a browser cookie. Cookies only tend to identify the final few interaction on a single device immediately before a customer converts. The nature of typical buying patterns dictate that these last few touchpoints are likely very likely to contain bottom-of-funnel marketing activity like Paid Search.
Compound that with a last-non-direct attribution model - which identifies any touchpoint which is not direct and gives it 100% attribution ignoring all other touchpoints - and you end up with a view of your data which contains as little Direct attribution as it is possible to orchestrate. Last-non-direct is a default model in a majority of other attribution services.
Corvidae takes a different approach. We use cookieless tracking methodologies to extend the fidelity of a journey to all devices and touchpoints across a much longer period, providing a view of your customer experience which is much truer to life. An ostensible side-effect of these additional touchpoints is that they are as likely to be Direct as any other marketing channel.
Our attribution models are also designed not to neglect touchpoints arbitrarily, but assess all interactions fairly giving weight to behaviours which indicate an increased conversion likelihood.
These factors together result in a greater propensity for Direct attribution. We mitigate that tendency somewhat by back-attributing Direct visits which were caused by previous visits, for example: If a user were to click a paid link in Google at lunch time, then later visit your site from the sofa by typing your website into their phone, we would attribute the value of the Direct sofa session back to the lunch time ad campaign.
What can I do to reduce the amount of Direct traffic in my performance data?
Mark up all marketing links with query parameters
It is standard practice in marketing that when a link is published query string parameters are appended as identifiers.
For example, if I would like to direct my social following on Instagram to my website corvidae.ai I might publish this link:
https://corvidae.ai/?utm_campaign=new-site-promo-2025&source=Instagram&medium=ppc
I now know anyone who clicks on that link have engaged with my new-site-promo-2025 campaign, came from Instagram, and came from ppc activity so I can determine that I paid for their visit.
This practice is not limited to paid activity. If I wanted to direct my YouTube subscribers to the same site using a link in a video description I might compose this URL:
https://corvidae.ai/?utm_campaign=attribution-is-the-way&source=YouTube&medium=organic
I can determine that these users engaged with my attribution-is-the-way video on YouTube and the click was organic. This last quality is important, because I might sometimes pay for the same video to to be displayed as an ad using this URL instead:
https://corvidae.ai/?utm_campaign=attribution-is-the-way-jan-prospecting&source=YouTube&medium=paid
I know these users engaged with the same video on YouTube, but this time via my attribution-is-the-way-jan-prospecting campaign which is part of my paid social strategy.
Once all published links are market up appropriately I can then set my Channel Assignments in Corvidae to allocate them to the desired channel for reporting and analysis.
Let Corvidae do the heavy lifting
We endeavor to make your life easy and your data useful. To avoid the need for this tedious manual tagging we intend to include the feature to automatically detect the origin and campaign of any Campaigns on connected sources.
Presently our Google Ads and Meta Ads connections can automatically detect this information without any input from you.
Know your customers
If you have repeat customers who visit you often by typing your website address into their browser or hitting a bookmark, you might naturally have a lot of Direct traffic. This is no bad thing. This sort of customer is both loyal to your brand, and cheap or free to re-acquire.
Upon identifying this customer segment there are strategies you can employ to better understand and further engage them:
Design an email strategy for loyal customers allowing you to identify the interest categories or product lines which separate them
Build remarketing cohorts and reengage them with products of interest while they surf
Either of these approaches will gradually introduce non-Direct visits to your typically Direct customer journeys and give you markers of interest on which you can act.
Choose the right model
If you have done everything above and the amount of Direct traffic in your reporting data still leaves you with questions, it might be that the lens you are using is the wrong choice for the customer segment being interrogated.
Attribution models have strengths and weaknesses, and no perfect solution currently exists. Some are great at identifying purchase intent (like our Customer Conversion Model) while others are better at identifying the first delightful interaction (our Customer Discovery Model) or the growth of brand awareness.
In long, considered journeys, the CDM is a good choice when interrogating top-of-funnel activity like social. Whereas the CCM is a better choice in bottom-of-funnel conversion activity like in high-intent search campaigns or “we’ve saved your basket” email targeting.