How to Read a Verification Result

Full transcript of How to Read a Didit Verification Result - Every Check Explained - Didit Academy video 5 of 10, 14:26. Watch on YouTube. Each paragraph links to the exact moment in the video.

​

What this video covers

00:03 Every verification you run in Didit comes back as a detailed result and that result is laid out as a row of tabs across the top of a session. In this video, we’re going to go walk through those tabs in order left to right so you know exactly what each check does and what it looks like when one of them fails. And by the end, you’ll be able to open any session and read it with confidence from the headline verdict right down to the evidence behind it. 00:30 So, let’s open one up and get stuck in.

​

Finding a session

00:33 Okay, so let’s navigate to the user verifications. This is where every session is going to appear from all of your workflows. It’s going to be one row per person. And normally you’d filter by status, workflow, or date to find whatever you need, but today we’re just going to go straight into a single result. So, I’m just going to click on one of these and open it. So, right here we can see this

​

The overview tab

00:55 is the overview tab. These are collapsible so we can collapse these. We can collapse all the checks if we want if we want to see everything, but honestly they’re all up here as well. So, you could see this particular user has passed all of these specific checks. So, if we just head over to overview, we

​

Warnings: where to look first

01:13 can pull that out and you can see we’ve got a few different tabs here. Now, the most important thing to check out here is the warnings. So, you could see this user is in review and you can see they’ve been flagged for a duplicate phone number. Okay, so if we check out other people that are in review, we could see the warnings in the overview section might look a little bit different. Based on the flags from the workflow, this particular user has gone through. You could see the database validation and the name mismatch. So, this is where we need to look if we’re reviewing a verification and we need to figure out what the issue is and then we can go to the specific section and we can see where these checks actually are and we can look at the data that’s been submitted to see if there is indeed a mismatch and what’s really going on there. Now, if we remove all of these filters, so I can just come out of here,

​

An approved session

02:04 we can remove these filters. So, I’m going to select all of these and let’s see if we go into someone who has been approved, we can see there’s there’s no major warnings that have been flagged. We can see this has been listed under the warnings, but it’s not triggered the in review or the decline. This has actually been approved. We can see this little pill up here and we can actually change these as needed. We’ll get into

​

A declined session

02:26 that in a little moment. And what does it look like when someone has been declined? We can see here the warnings, again, the name mismatch. And then up here at the top, we can see proof of address is the one that caused some issues. So, we can come in and take a look at this as well. So, we’re going to go through each section one by one so you could see exactly what’s happening underneath the hood. So, I’m just clicking into this profile here and you can see the liveness and the face match is where this particular user has been flagged for checking. So, we’ll get to that in a moment. But the rest of the

​

Session details, vendor data and cost breakdown

03:01 tabs in the overview section, we’ve got session details, so you’ve got the session ID, when it was created, the vendor data. So, we spoke about this in previous videos. The vendor data is a unique identifier for this particular user. This is where all of the [snorts] checks, all of the verifications and transactions are going to be collected under the specific vendor data. We can see the workflows that this user has gone through for the verification. We’ve got the contact details here and any tags. I mean, we can we can add a tag here if you want. 03:31 We can tag people in workflows as well. Now, the total cost breakdown, I believe this this user was imported. You can see here on the left, it was imported, so he hasn’t actually gone through a workflow. Let’s click this other user here and you can see the workflow that they’ve gone through and you can see the cost

​

ID Verification: authenticity before data

03:48 breakdown of all the checks this user has completed. So, let’s take a look at the ID verification. Now, the main job is not just reading text off the card. It’s It’s proving the document itself is genuine before anyone trusts a single field on it. So, under the hood, we’re running tamper detection, document liveness to catch a screen replay or a printed copy, security feature validation for holograms and watermarks, and a a cross-check between the printed fields and the machine-readable zone, the MRZ, the block of chevron characters at the back that you could see here on this particular ID. Now, once the document is proven authentic, Dedit will read the fields with OCR and lay them out. Here, you can see the personal data. Got the document type, the document number, personal number, issuing state, all of the information that you would find on the ID, including the address data and any additional data in here. And remember, Dedit can read more than 130 languages across 14,000 document types. So, you’re going to get clean fields

​

The checks list

04:47 wherever someone presents their ID. So, there is a tab here called checks list. And these are the checks that are running in the background on this specific ID. So, I’m not going to read through all of these. There is a lot here, but this is essentially how Dedit is verifying that this is an authentic ID. So, you don’t have to configure any

​

Liveness: score, video and quality

05:07 of this. This is all running in the background. So, the next check is the liveness check. You can see that this particular check is in review. This is what’s been flagged. You can see liveness and face match. Now, I will say this is a demo account, so some of this data is not matching. Um you can see the liveness score is it’s not 100%, and if we play the liveness video, you could see it’s it’s not actually the person. 05:30 This is just a random example. Now, if you were to see this on a on a real user, you would see the the person doing the liveness check, the video, whether it’s passive or active. We’ve got the face quality score and the face luminance score. And the cool thing about Dedit is if you’re confused about any of these scores, you can just click this button here and it’ll give you an explanation of what this does. We can change the status of any of these checks if we believe that this has been flagged inappropriately, and obviously we can change the overall status of this verification. Again, just like the ID verification, we’ve got a checks list and there’s a bunch of checks here that are all happening in the background, right? And we can hover over these and we can see a little bit more information about each of these checks. And if somebody fails any of these tests, instead of the green

​

Face Match

06:16 tick, there’s going to be a red X. So, if we keep going, we’ve got the face match tab. This is going to match the face on the ID with the face in the liveness check. And you can see here, obviously the similarity score is pretty low because they’re not really the same person. And again, we’ve got a checks list of what’s happening in the background there and what this particular check is doing. So, the similarity score is based on the original ID and the liveness image that we’ve captured. And this is to figure out if the person holding the ID is a genuine owner. Now, we’re not going to see this in the dashboard, but there is a face

​

Face Search and block lists

06:54 search. So, beyond matching against the document, Didit searches the same face across all of your other sessions. So, this part only shows when it finds a match. So, if the same face turns up in another session, a grid of matches is going to appear here. And if one of those matched faces is on a block list that you maintain, it’s going to be flagged and the session will be declined, which is how a known fraudster gets stopped on site. But on a clean session with no matches, this area stays

​

Proof of Address

07:21 empty, so we’re not going to see anything here. Let’s move on to proof of address. We can see the document that was submitted and the data that was extracted from this particular document with all of the metadata. So, we’ve got a checks list here and we can see one of the checks does not have a green tick, meaning that this check was being flagged. And we can see one is in review. Now, that’s not enough for this proof of address module to be flicked over to declined, but this is again something that your human reviewer is going to be looking at when they come to a profile that is in review. And a pass here means the code was entered correctly and a failure means the number couldn’t be confirmed, and you’ll see that reflected in this module here. So, you remember at the top, one of the flags here was duplicate phone number, and now that’s probably due to this being a demo

​

Phone verification and duplicates

08:08 account, but you could see when we scroll down under the phone verification, we could see phone matches. Now, this is going to search the phone number on your database, and if it matches anything from a previous user, then this is going to be flagged in this tab here. We can also hover over here. We can see other verification sessions that use the same phone number. This is just an explanation of what this module does.

​

Email verification

08:35 Same for email verification, the user is going to receive a code and put the

​

AML screening: sanctions, PEPs and adverse media

08:40 code, and if everything’s good, then this module is also approved. Now, the anti-money laundering, the AML screening, there’s quite a lot going on here. This checks the verified person against sanction lists, politically exposed persons, adverse media, and then reports how many matches were found. So, you could see the data that it’s using to screen, and we can actually see the list. So, if I open this up, we could see all of the sanction lists here that we can run the checks against. So, narcotics, terrorism, financial crime, everything. And we can see if this user has matched on any of the lists, we could see here the match score, the risk score, and the lists that this user has appeared on and whether it’s a false positive or not. Now, a false positive can occur if the name matches somebody on one of those lists. So, this is where we have other information like the date of birth to to cross-reference. We do have a checks list here again. Uh it’s pretty short for this particular one, but the the checks against all these lists is doing the heavy lifting here. 09:41 And if you want to have this person monitored ongoing, where we can check the lists against this user on an ongoing basis, we can actually turn that on. If we keep moving down, we can see

​

Device and IP analysis

09:50 the device IP analysis. So, we have the ID verification document location. So, the actual address from the document. We have the address details here for the proof of address. And then we have the device information. So, the IP address and where that person is actually completing the verification from. And we could see at the bottom the document location versus the proof of address, how far away those distances and those locations are from the verification IP. And we can see all of that on this map. So, we see if we hover over here, we’ve got the ID verification document location. 10:28 Got device number one, so they’re pretty close. And then we have the proof of address in the US here. Again, this is a demo account, so these details are not particularly accurate, but this just gives you a good understanding of is that person close to where the proof of address and their ID actually is. And here we have the checks list again, and these are all the checks that ID it is doing in the background to decide whether to flag [snorts] this particular check or not. If we keep going down, we’ve got database validation. So, if you remember in previous videos, we spoke about database validation. It’s a node that you can add in a workflow. And what we can do here is we can run a check. And for this example, we can run

​

Database validation: full, partial and no match

11:06 a check in Argentina, and we can check against a government registry, civil registry, the tax authority, the electoral roll, all of these different directories that we do have access to. You could see we have a match found here. And you could see the source data, the fields that match on the results. So, we’ve got a full name match and a date of birth match. And if we open up the checks list, we can see one of these checks was flagged. 11:29 And we can see here the match type. Now, if a full match has been found, it means the identity exists in the registry. Partial match means some fields matched and you should review the rest. And a no match is a strong synthetic identity signal because a fabricated ID number simply isn’t in any registry. And this is one of the strongest defenses you have against made-up identities. And again, it runs in the background with no

​

Document AI and questionnaires

11:54 extra steps for the user. Now, moving on, we do have a few other tabs here and these can vary depending on the checks that this particular user has gone through. We can see document AI. This check hasn’t run. You can see not started. There are other checks that we do have for questionnaires. If somebody ends up filling out a questionnaire, we could see the answers to those questions in

​

The event log

12:15 the questionnaire tab. And then if we keep on moving down, we’ve got the events. So, this is a audit trail, a log of all of the events that happened. So, when the session was created, when the session status changed, and each entry carries the device, the IP, the timing, who did it, so you can reconstruct

​

Webhook deliveries

12:31 exactly how a session unfolded. And then finally, at the bottom, we have webhooks. So, webhooks list every delivery Didit attempt to your own endpoint with the response status. We can see 200 here. And we could see the request header, the body, the webhook URL, the status of this particular verification. And we’ve also got the date here this webhook was sent. So, one thing to note here is that this dashboard can be accessed through the API through the MCP. You can ask Claude to pull all of this data, to pull all of your verifications that are in review that you can analyze inside of

​

Resubmission, PDF export and block lists

13:08 Claude as well. Now, if you take a look at the top here, we’ve got some breadcrumbs and we can see we can request resubmission. We can download the PDF, which is excellent if you want to have an audit trail or you need to present something in a meeting. Um you can also access this download through the API and you can add this person to a block list. So, that is how you read a verification

​

Session chat for your review team

13:30 result. There’s one last thing at the bottom here. We’ve got a session chat. So, we can actually tag team members, and we can talk about a specific verification result with somebody on our team. So, if we opened up this chat and we put in a message and we tagged someone on our team, that’s going to be specific for this particular verification. So, if we do have a question or there’s multiple people working on this case, this chat is going to be incredibly useful. 13:55 So, that’s a verification result read the way you’ll actually read it, tab by tab, top to bottom. The verdict and the warnings on overview, then each check in turn, and the full audit trail at the end. So, once you’ve done it on one session, every other session reads the same way. And if you want to try it on your own data, head to didit.me, sign up for free, and open your first result.