Re: An introduction to Highstreett - An update on how Highstreett search works as of October 9th 2026

#Some things have changed#

Hello,

I have made some product updates since my introduction blog a couple months ago. Highstreett no longer just pulls independents it instead filters to prioritise them using a series of negative signals. When pulling from areas we have covered it pulls from the database then when pulling from a new city in the UK or abroad we leverage google places then log the businesses in the database afterwards.

Using the UK as an example and where the results come from:

Where results come from

Every search asks Google Places (up to 20 results) and our own listings in that town and towns within about 10 miles. If both return the same place, our listing wins. It works in any UK town, even ones with no city page. There’s no “open web” fallback for finding places. Lincoln is deepest: 437 published listings, 33 verified independent. London is next with 48. Once a business appears it is logged into the DB this is to save on API costs and as we can build better context/do better verification on the businesses. This is our independent badge is awarded but we do our best to provide you with more personalised local picks even in cities without DB stored businesses.

What is the result shown?

Google ratings with the review count. Opening hours, directions on how to get there then for businesses such as restaurants you will be able to see their menu. Highstreett ratings will be displayed alongside google reviews with the intended goal of eventually replacing them.

Ownership

Three labels: verified independent (checked by hand against Companies House), chain or pub-company or private-equity owned, or not checked yet. Most places from Google start as “not checked yet”. A short list of big national chains is left out entirely: fast-food giants, big coffee chains, supermarkets. A background check can mark a place as a chain, but it can never give the badge. Only a person does that. Shop Local hides chains and keeps unchecked places, labelled as unchecked.

We only do the verification for businesses in the DB as the check is ran independently after.

Result ranking?

We fit places/books out of 100 with the indented goal matching places most accurately towards your query. We aim for fit more than straight review rankings although reviews can be a deciding factor this is so that smaller and newer establishments are not just overshadowed like seen with google places.

Signed out vs signed in

Signed out:

it reads your request (type of place, area, budget, mood, dietary needs) and remembers what it’s shown you in that chat. There’s no profile.

Signed in:

Your ratings, reviews and likes move places we list up or down by up to 15 of the 100 points. A pattern carries over too: love small bakeries and other small bakeries rise. Saved dietary needs, allergies, religious requirements, dealbreakers and notes go with every search. It remembers across chats what it recommended and your feedback from the card buttons: Liked it, Not for me, Already knew it, Been there.

Highstreett is built on personalisation so the more you use it the better the product gets.

How the directory grows?

Either our manual city roll out where we curate it using the highstreett search and list them so they can be found in discover and not just search or by user usage. Search is the engine where you get personalised recommendations and discover is when you want to window shop or browse by category ie look through a specific book stores stock.

Limitations?

Highstreett is slower than both google and chatgpt by a few seconds sometimes 5-10 seconds per result.

We are also still in our discovery/directory phase so you cant transact directly through highstreett instead we navigate you to go to the business in person. This makes us a more personalised Yell/Yelp rather than a full marketplace however that is our intended goal.

I will come back and do another update blog when we make further product improvements to the search and marketplace/directory

Thank you,

Alex