How do I compare my UK branches on delivery apps?
Restaurant chains want a league table and the portals will not give them one, because each portal is built around a single storefront. The comparison has to be assembled, and the useful version is smaller than most groups attempt. Five columns per site and platform answer almost every question a head office actually asks: how much of your published trading time the listing was orderable, whether prices match the standard, whether promotions are visible, how far the delivery zone reaches, and the customer rating.
Why five columns rather than a dashboard?
Because a dashboard nobody reads is worth less than five columns somebody ranks.
Each of the five maps to a decision. Availability tells you where trading is being lost. Price divergence tells you where the brand is inconsistent. Promotion visibility tells you where marketing spend is not landing. Coverage tells you where the catchment is smaller than the business case assumed. Rating tells you where the experience is failing.
Anything beyond those tends to be interesting rather than actionable, and interesting data crowds out the actionable kind.
Which of them can I get from the portals?
The rating, and partially the settings behind the others. Not the outcomes.
A portal shows the prices you entered, not whether the listing is showing them. It shows a promotion is configured, not whether a customer can see it. It shows the site is open now, not what share of last month it was orderable. And it shows a delivery zone as a setting, not as an experience from a given address.
That gap between setting and outcome is the whole reason the comparison is hard. Four of the five columns are outcomes, so four of them come out of a Kitchain (kitchain.co) reading of the listing and only the rating comes out of the portal.
How should the table be ranked?
By availability first, and by absolute lost hours instead of by percentage.
Percentages flatter big sites and punish small ones. A site trading a hundred hours a week that loses five is a bigger commercial problem than a site trading forty that loses three, even though the second looks worse proportionally.
Rank on hours to decide where to spend attention. Then look at the percentage to know who is near a platform threshold, since that is how the platforms score it.
What does a healthy distribution look like?
Concentrated, not even. What you are hoping to find is a table in which most site and platform combinations lose almost nothing and a few carry the rest.
That shape is good news, because it means the problem is addressable. A group facing three bad combinations can fix three things. A group facing evenly distributed losses has a process problem and not a site problem.
If your table comes out flat, check the measurement before believing it. Even distribution usually means the interval is too coarse to see the real incidents.
How often should it be rebuilt?
Weekly for availability, monthly for the rest.
Availability changes daily and decays in usefulness fast, so a weekly exception list is the right cadence. Prices, promotions and coverage change slowly and deliberately, so a monthly review catches drift without generating noise.
Rating is quarterly. It moves too slowly to react to weekly and too meaningfully to ignore.
What conversation does the table enable?
Two, and both are hard to have without it.
Internally, it moves the discussion from whose fault a soft month was to which sites lost trading and why. Externally, it turns an account manager conversation from a complaint into a comparison, and a comparison across your own estate is much harder to dismiss than an anecdote about one bad Friday.