We read the public record. Then we print our working.
There is no panel of experts and no proprietary index. Five public sources, swept every
Monday for your shop and the four you name, put through rules you can read on this page,
and every figure opens into the sentences it was computed from.
0public sources, read in full
Monthe sweep runs, every week
0figures produced by a language model
Five steps, and you are present for four of them.
The only part you do not see is the sweep itself, and the console tells you when it ran
and what it could not reach.
01
You name your shop
Your restaurant's name and locality, typed the way a guest would type it. We match it to
your live Google listing and show you which one we found before anything else happens.
A listing matched to the wrong outlet poisons everything downstream, so you confirm it.
you get your listing, scored out of 100, in about two minutes
02
You name up to four rivals
You pick the four doors you actually lose customers to, rather than letting an
algorithm guess at your competitive set. They are read with the identical sweep, in the same week, from the same
sources. Change the set whenever the street changes.
you get one shared axis with the four shops that matter
03
The sweep runs every Monday
Google, Zomato, Swiggy, Reddit and Instagram, read in full for five venues. Reviews,
ratings, hours, photos, menus, topic chips, local threads and posts. Weekly, not
real-time: a restaurant's public record does not change by the minute, and an alert that
fires on Tuesday afternoon finds nobody free to act on it.
you get a fresh week, dated, with the run time printed
04
Rules rank what to do
Every recommendation is fired by a rule with a threshold: an unanswered review count, a
listing field that disagrees with reality, a dish whose mentions have turned, a rival's
pace pulling away from yours. Ranked by what the thing is costing you, with the minutes
it takes printed next to it.
you get a ranked list, not a feed
05
You check our working
Tap any figure and the sentences it was computed from open underneath it, with the
source mark, the rating and how long ago. If we could not reach something, the panel
that would have shown it names the gap instead.
you get a claim you can verify in one tap, or reject
Five sources, and what each one actually gives us.
The second column matters more than the first. A tool that lists five logos and does not
say where each one stops is asking you to guess how much of your week it has read.
SourceWhat we readWhere it stops
GoogleReviews and ratings in full, replies, opening hours, photos, the menu where it is listed, and the topic chips Google itself computes.Exact review dates. Google reports “a month ago” and means it, so we never show a day or a time derived from one.
ZomatoPublic listing, rating, reviews, dish mentions, the menu as published, and delivery-versus-dining signals where they are shown.Anything inside your partner dashboard: covers, order volume, ad spend. That is yours and behind your login.
SwiggyListing signals, menu and pricing as published, rating, and the review text where the listing exposes it.Order-level data, and any listing that has been pulled. When a rival's listing disappears mid-week, the console says so.
RedditLocal threads mentioning your venue, filtered to your city so a name collision in another country is not counted as a mention of you.Private subreddits, deleted comments, and DMs. Volumes here are small by nature; a quiet week is reported as quiet.
InstagramYour public grid and your rivals': posting pace, what gets engagement, and what the captions and comments name.Stories, DMs, and anything on a private account. Rival posts are read in your timezone, never theirs.
All five are public surfaces, read the way any guest could read them. Connecting your own
Google or Instagram account is optional, adds your own first-party numbers, and can be
disconnected at any time without losing the weekly sweep.
Five rules, enforced in code, not in a promise.
These are not values on an about page. Each one is a constraint the product is built
around, and each one costs us something, which is how you know it is real.
01
Nothing is invented
Every figure opens into the guest sentences it was computed from. A field the sweep
could not answer arrives empty, and the block that would have shown it disappears.
Never a zero standing in for a gap, never a dash, never a blurred teaser number behind
a paywall.
costs us A page with holes in it, on the weeks the sweep came back thin.
02
No causation claims
Outcome tracking says “since you did this”, never “because you did this”. We can see
that you fixed your hours on the second and that nobody has complained about a closed
door since. We cannot see why, and we will not model it.
costs us The screenshot every competitor ships: “+0.3★ attributed to Outstrip”.
03
Relative dates stay buckets
Google says “a month ago” and means a range. Nothing in the product displays a precise
date or a time of day derived from one. Weeks are compared as weeks, not as timestamps
we sharpened ourselves.
costs us A tidy timeline. We would have had to invent the days to draw one.
04
A dish is named only if your menu backs it
Guests misname things, and a text match will happily report enthusiasm for a dish you
do not sell. Every named dish is matched against the menu we can actually see on your
listings; the ones that do not match are dropped rather than guessed at.
costs us A longer, more impressive dish list than the true one.
05
Rules fire the recommendations, not a model
Every action in your week comes from a rule with a threshold you could check by hand.
The AI writes the sentence around it and drafts the reply. It never produces a figure,
and it never decides what matters.
costs us Surprise. A rules engine cannot dazzle you with a finding nobody asked for.
What a public sweep cannot see.
Printed here for the same reason it is printed in the console: naming the gaps is what
makes the rest believable. If any of these is the number you need, Outstrip is the wrong
purchase and we would rather you knew now.
Covers, revenue and your POSbehind your login
WhatsApp and phone ordersnot public
The exact day a review was leftGoogle says “a month ago”
Why a number movedwe say “since”, never “because”
A rival's real prices after discountlisting price only
Anything said on a private accountout of reach, by design
Where the AI is, and where it is not.
Worth being exact about, because “AI-powered” usually means nobody can tell you which
number came from a model.
The AI writes
It turns a fired rule into a sentence an owner can read at the pass, drafts the reply to
a review in your register, and groups sentences that are making the same complaint in
different words.
Language only
The AI does not count
It produces no figure on any screen. Ratings, mention counts, reply rates, pace,
rankings and thresholds are computed arithmetically from the swept records and can be
checked by opening the sentences underneath.
Every number is arithmetic on a record you can read
Your data, in four sentences.
What do you store about my restaurant?
The public records we swept (reviews, ratings, listing fields, menus, posts) plus
what we computed from them and which actions you marked done. Your account holds your
name, email, phone and the outlets you added.
Do you post anything on my behalf?
No. Replies are drafted for you to edit and send. If you connect Google, sending still
takes your explicit tap on each one. There is no auto-post setting, because there is
no version of it we would be comfortable shipping.
Is my data used for anyone else's console?
Your own first-party numbers, no. The public record about your restaurant is public,
and it can appear in the competitive set of an owner who names you as a rival, exactly
as theirs appears in yours, and exactly as either of you could read it on Google today.
What happens when I cancel?
The sweep stops at the end of the month you have paid for. Ask us and your data is
deleted; say nothing and it is kept for ninety days in case you come back, then
deleted.