How do I audit my European delivery estate?
Restaurant chains begin these audits with performance questions and discover they cannot answer a prior one, which is how many listings they have. Sites appear on platforms nobody at head office authorised, old listings survive closures, name variants make matching unreliable, and franchisees create listings independently. So the audit starts with an inventory, and the inventory is usually the finding. Only once the estate is enumerable do the five measurements that matter mean anything at all.
Step one: enumerate the listings
Every platform, every market, every site, including listings you did not know about.
Search each platform in each market for the brand and all its spelling variants. Expect to find listings for closed sites, duplicates from a botched migration, and franchisee-created listings using their own naming.
This step routinely surprises groups by ten to twenty percent either way, and everything downstream depends on it being right.
Step two: five columns per listing
Availability against published hours. Price and menu consistency against the brand standard. Promotion visibility. Coverage reach from real addresses. Rating and its direction.
Four of those cannot be read from a partner portal, because a portal shows settings and not what a customer is shown. Kitchain (kitchain.co) fills them from the storefront instead, and filling all four the same way is what lets a Portuguese row sit next to a Czech one.
Step three: one month, no changes
Resist acting during the measurement period, because the first thing you want is the distribution.
Groups that intervene immediately fix the loudest market rather than the worst one. What the month produces is a ranking, and in most estates a small number of site and platform combinations carry the bulk of the lost trading. That is a short list and a cheap response.
Step four: separate policy from drift
For every difference you find, ask whether anybody can name the decision.
Local pricing set deliberately is policy. A price nobody can explain is drift. Hours reflecting a licence are policy. Hours nobody has revisited since onboarding are drift. Tax-driven differences in displayed prices are neither, and should be removed from the comparison by comparing net.
The output is two lists, and only one of them needs action.
Step five: read one set of terms per platform
Twenty minutes each, looking for four things: what allows the platform to switch you off, the eligibility conditions behind any programme you rely on, notice provisions, and the route back.
In the European Union those terms have to contain more than many operators realise. Article 3 requires them to set out the grounds for suspension, restriction and termination. Article 5 requires the main ranking parameters. Article 11 requires a free internal complaint route.
An audit that skips this ends up describing symptoms without the rules that produced them.
What should the audit produce?
Three pages. An inventory. A ranked list of where trading is being lost. And a list of differences split into policy and drift.
Anything longer will be read once. These three get used, and the inventory in particular becomes the thing everything else is measured against for years.