The most expensive mistake we see estates make after an Irisys Gazelle or Vector 4D fails is not the hardware decision — it is the assumption that a dead sensor means a dead platform. Since Irisys hardware reached end of life in March 2023, failed units cannot be replaced like for like, and that single fact has pushed a lot of retailers into tender processes for entire new counting ecosystems they never actually needed. The sensor and the platform are separable problems. Treat them that way and an irisys sensor replacement becomes a rolling maintenance line item instead of a capital project.
When a Gazelle unit above a busy entrance stops reporting, the pressure is immediate. Conversion rates go blind for that door, footfall trends break, and staffing models built on entrance data start drifting. The instinct is to call whoever supplied the original estate, and the answer that often comes back is a proposal to rip and replace: new sensors everywhere, new head-end software, new integrations into your BI stack, and a data history that either gets migrated at cost or quietly abandoned.
That answer serves the vendor more than it serves you. If your platform is sensor-independent, a failed Vector 4D is just a device swap: one ladder, one new unit on the same mounting point, one configuration session, and the data stream continues under the same store ID with the same historical baseline. The rest of your estate — the hundreds of Irisys units still working perfectly — carries on untouched.
Two realistic candidates dominate the shortlist for defective-unit replacement, and they solve slightly different problems.
On the Vemco platform both run alongside your remaining Irisys estate, together with Elsys, Hikvision and Axis devices where those already exist or where a camera-based approach makes sense for a specific site. Mixed estates on one platform are not a workaround — they are the operating model that makes gradual replacement possible at all.
Run the numbers on a 200-store estate. A forced single migration means 200 sensor purchases, 200 installation visits, and a coordinated cutover — all in one budget year, all before a single unit has actually failed. A gradual approach means you replace on failure or on planned refurbishment, spreading spend across three to five financial years while the working Irisys units keep earning their original investment back.
There is a second, less obvious saving: installation logistics. When a sensor swap coincides with a store refit or a scheduled maintenance visit, the marginal cost of the swap drops sharply because the engineer, the access equipment and the out-of-hours window are already paid for. Estates that hold a small buffer stock of Milesight units — even four or five devices — can fold replacements into whatever site visits are already happening rather than dispatching engineers reactively at emergency rates.
Footfall data is only as valuable as its comparability. A new sensor that counts differently from its predecessor quietly corrupts every year-on-year comparison for that store. This is where discipline matters more than technology.
Here is the practitioner detail that rarely makes it into vendor brochures: when you swap a Vector 4D for a Milesight unit, do not assume the old count line placement transfers. The two sensors see the entrance differently, and a count line drawn in the same nominal position can capture or exclude different edge traffic — staff cutting across the vestibule, people lingering in automatic-door zones. The reliable method is a validation count after installation: a manual tally against sensor output across representative trading hours, adjusting line position and exclusion zones until the new unit's behaviour matches the accuracy standard, not the old unit's quirks. Skip this step and you can end up with a sensor that is technically accurate but statistically discontinuous with your history.
Accuracy itself should be contractual, not aspirational. Vemco commits to a minimum of 96 percent counting accuracy, and in practice typically achieves 98 to 99 percent where conditions — lighting, store layout, visitor behaviour — allow. Be wary of any supplier quoting a flat 99 percent with no conditions attached; entrance environments vary too much for that to be an honest blanket claim.
A workable rollout order, based on how these programmes actually go:
Migrating away from Irisys hardware goes better with a partner who knows it intimately. Vemco Group, founded in Denmark in 2005, has been an Irisys partner for more than 15 years — which means the people configuring your Milesight or Xovis replacements understand exactly how the old Gazelle and Vector units were set up, where their count lines sat, and what behavioural quirks the historical data contains. That institutional knowledge is what turns a sensor swap into a continuous dataset rather than a break in your records.
If you have failed Gazelle or Vector 4D units today, or you are budgeting a sensor refresh for the coming financial year, talk to us before committing to a full estate replacement. We will map your current Irisys estate, recommend the right successor sensor per site, and show you how a mixed estate runs on one platform while you migrate at your own pace. Get in touch at vemcogroup.com/contact-us and bring your device list — we will tell you honestly which units need replacing now and which can wait.