If your Sumsub account only checks documents and selfies, put Veriff and Didit on the list. Persona makes sense when the team needs to route inquiries and review cases. For registry and ownership data, compare Trulioo. Price transaction monitoring separately if your current account handles it too.
Give each candidate a support ticket involving an applicant who is stuck in review. Ask the team to recover the result and show the decision record it could hand over after a switch. You'll learn more about the day-to-day work than you will from the capture demo.
For an ID-and-selfie replacement, we'd shortlist Didit at 7/10. I'd try its hosted flow with a support analyst in the room. The capture and individual checks fit a small IT team's brief. I'd reserve the remaining points for a clean retry, an evidence export and a clear handoff if the current account also uses Sumsub screening. If the analyst needs a developer to locate the result, I'd keep the current setup for that part of the workflow.
Identify what Sumsub does in your account today
Start with a sample approved account and a sample unresolved account. Find the applicant, the checks used, the review status, any alert still active and the history that supports the decision. Ask which parts of that record can be exported and who can see each one. A procurement inventory that lists “KYC” as a single feature misses whether your operators depend on proof of address, ongoing screening, company verification or transfer-related rules.
Sumsub documents screening, business verification and transaction monitoring separately. Its pricing page also separates entry packages from custom features. Check which modules you use before requesting a quote. Swapping document and face checks is a smaller job than moving a compliance programme with several teams.
Look at the vendor bill, the onboarding settings and the support queue together. A feature might be enabled without being used. A recurring task may exist outside the provider even though the sales document calls the package end to end. Have the person who handles exceptions mark each capability as critical, optional or obsolete.
Which alternative fits each kind of replacement?
| Alternative | Good candidate for | Run this test before accepting it |
|---|---|---|
| Didit | Replacing selected identity steps with a hosted workflow or standalone checks | Carry one pending case through retry and confirm you can keep the same internal customer record |
| Veriff | A document-centric flow with automated decisions or a plan that adds assisted review | Compare the exact retry reasons, evidence export and support access in the chosen plan |
| Persona | Configurable inquiry and operational decision paths | Show the difference between user completion, review and approval in your own support console |
| Trulioo | Business verification where entity and ownership sources are decisive | Compare country and entity type coverage against your real customer mix |
| Keep Sumsub for some modules | An active screening or transaction workflow that need not change with capture | Prove how the two vendors' results and incidents are connected without duplicating review |
Didit documents hosted capture and calls to individual checks. A hosted session takes camera work off a team without a dedicated mobile engineer. Standalone calls fit a team that already has approved capture and storage.
Veriff's public pricing describes plans with different review and export features. Test the plan you will purchase, especially if staff must retrieve a PDF or have a human look at a case. Persona's docs show hosted and embedded paths. A flexible inquiry can suit different client organizations, but every custom branch needs an owner and a support response. Trulioo documents a KYB API; ask for coverage at the registry and legal-entity level rather than assuming all businesses in a supported country receive the same result.
Put the service desk in the pilot
Give a support analyst a fictional ticket: “My photo was accepted, but the account still says pending.” Can they locate the account using an internal reference, see the current provider attempt and tell whether the decision is waiting on a reviewer? Can they give the next step without seeing a copy of the ID? If the test needs access to a developer's private logs, the new operating model is incomplete.
Next, send an incident to the engineer: the provider says the case is finished but the application never recorded it. Check whether there is a result endpoint or another documented recovery path. When the result is reprocessed, it should update the existing case once. Use an old duplicate event to verify it does not overwrite a later decision. Time this recovery separately from the happy-path capture.
Finally, have the policy owner inspect an evidence export. A screenshot of a green badge is rarely enough to know which workflow ran. An operational record should be able to show the provider reference, the decision, the policy version, review activity and relevant dates. Exact retention and access are choices your organization has to make. Ask who can retrieve and remove evidence on each side of the integration.
Replace by layer when the whole stack is not broken
If your main issue is a confusing camera step, it may make sense to trial a new capture flow while retaining other modules. When your problem is that three consoles hold parts of the same customer decision, adding another provider could make the issue worse. Draw a boundary around the workflow you want to change and write down how the other parts receive a result.
For example, use a limited cohort for a new identity workflow and keep the existing records where they are. Give the customer a single path, not two requests to upload the same document. Complete cases that were already pending in the original system and decide in advance when a genuinely new attempt should go to the replacement. Review rollback as a routing decision for new attempts, with separate handling for cases already in flight.
Use the small-business IT health check to assign owners for alerts, credentials and service recovery. The identity verification comparison for small IT teams can help narrow the initial list; the difference here is that you must carry over a process that already has real customers and an audit history.
Measure how many consoles the change creates
Ask one analyst to handle a case while narrating each system they open. If the case crosses your CRM, the capture provider, a screening console and a ticketing tool, record which reference connects them. A replacement that removes one contract might still add a separate queue. Test whether the analyst can assign the whole case to a colleague with one clear handoff. If not, fix the record linkage before assuming the help desk will compensate for it every day.
Budget for the replacement you are actually making
Compare the monthly minimum, volume, exceptions and add-on fees, but also estimate the hours to operate two systems during the cutover. The new provider may reduce the charge per check and still cost more for a quarter if analysts must reconcile duplicate records or you need to write a custom export tool. Include a month with unusually low volume and a week when the review queue is unusually high.
Run a normal case and an evidence export with two candidates, then give each the same retry and review. Keep the current setup if switching costs more than the work it saves. If you move, tell the help desk which provider owns cases that are already open.