The promise is lovely: type a postcode, pick your address from a list, done. Six fields become one. For a shopper in a standard house on an established street, that promise is kept, and it is genuinely faster.
Then there is everyone else. The flat conversion, the new build estate, the unit on a business park. For them lookup is not a shortcut. It is a locked door with their own address on the other side of it.
Know the three classic failure groups
New builds first: database updates lag construction, sometimes by months. A shopper who moved into a new estate this spring may not exist in the lookup data at all, and no amount of retyping fixes an absent record.
Flats and subdivisions second: lookup often returns the building, not the flat, or buries flat numbers in a list sorted in an order no human would choose. Business addresses third: units, trading estates, and care of addresses routinely return partial results that drop the company or unit line the courier actually needs.
The manual fallback is the feature
Every lookup deployment needs a visible enter address manually option, present from the start, not revealed after two failed searches like a consolation prize. When shoppers find it themselves, they use it without resentment. When they have to fail first, the failure is what they remember.
And the manual form must be complete: address lines, town, county or region where relevant, postcode. A fallback that still forces a lookup selection to proceed is not a fallback. It is the same door painted a different colour.
Never overwrite shopper input silently
The ugliest bug in this category: a shopper picks an almost right address, edits a line to fix the flat number, and the lookup revalidates on blur and restores the original. We have watched session recordings of shoppers having this fight three times before abandoning.
Rule: once a shopper edits a returned field, their text wins. Validate format if you must, suggest if you like, but do not replace. It is their address. They live there. The database, demonstrably, does not.
Measure lookup and fallback separately
Instrument both paths. What share of shoppers start lookup, what share finish with it, what share bail to manual, and what is completion in each group. A lookup tool with a 90 percent start rate and a miserable manual path can still be a net loss if the 10 percent it fails are effectively locked out.
Check delivery failure and courier correction fees by path too. If lookup addresses generate more corrections, often the dropped flat or unit line, the speed saving at checkout is being repaid, with interest, at the doorstep.
Related reading
These earlier guides go deeper on connected questions: Checkout Gift Options and Delivery Dates: Optimizing the Extras, Checkout Address Autocomplete: Fixing the Last Form Friction, and Reducing Checkout Steps Without Losing the Data You Need.
Frequently asked questions
Why is my address not in postcode lookup?
New builds, recent flat conversions, and some business addresses lag or miss database updates. It is a data coverage problem, not a typing problem. A good checkout offers full manual entry so a missing record never blocks an order.
Should I remove address lookup from checkout?
Not necessarily. It genuinely speeds up entry for standard addresses. The fix is a strong manual fallback, visible from the start, and a rule that shopper edits are never overwritten by the lookup result.
What is the biggest lookup implementation mistake?
Forcing shoppers to fail lookup before revealing manual entry, and silently restoring lookup data after a shopper edits it. Both turn a convenience feature into an obstacle for exactly the addresses the database handles worst.
How do I know if lookup is helping my store?
Track start, completion, and manual fallback rates separately, plus courier correction fees by path. Lookup only earns its place if overall completion and delivery accuracy both hold up, including for the shoppers it cannot find.