You are sitting on the research everyone else is guessing at.

DataPress reads the data your security team approves, finds the patterns worth writing about, and drafts them with every claim traced back to the query that produced it. Nothing publishes without a named reviewer signing it off.

See a story feedUses a sample company

Aggregate only. Read-only. Your scope, not ours.

This week, 4 stories

Harlow Freight · sample

Story 1 · lead

Refrigerated freight bookings in the Midwest, year on year, against 31% across all other categories.

━ refrigerated, Midwest━ all other categoriescohort 1,840 accounts

cohort 1,840 accounts · 18 months · 8 states · new accounts excluded

q_0412 Open the query

3.1×Customers who book through the API are 3.1 times more likely to still be active at 12 months than those who book through the dashboard.
1.6 daysAverage time between quote and booking has fallen from 4.2 days to 1.6 days over two years.
no figureSuppressed. Cohort 31, below floor of 50.

3 ready to draft. 1 suppressed, cohort too small.

Every claim has a query behind it

Click any number. The query is right there.

For immediate release · Harlow Freight · 19 October 2026

Midwest refrigerated freight bookings up 240% year on year, Harlow Freight data shows

CHICAGO. Bookings for refrigerated freight originating in the Midwest have risen 240% year on year¹ on the Harlow Freight platform, according to aggregate booking data released today by the company. Over the same period, bookings across all other freight categories grew 31%².

The two trends tracked each other closely until early this year. The gap opened in March 2026 and has widened in every month since.³ By September, refrigerated loads made up 18% of Midwest bookings, up from 7% a year earlier⁴.

The figures describe what Harlow Freight customers book on its own platform. They do not describe the wider freight market, and the company is not making a claim about refrigerated volumes outside its customer base. An industry-wide comparison would need a source Harlow does not hold, and none is offered here.

“Our customers are telling us something through what they book,” said Nadia Okonjo, Director of Communications at Harlow Freight. “We wanted to publish it with the method attached, so that anyone can check how we got there.”

All figures are aggregates drawn from groups of at least fifty accounts. Accounts opened in the last ninety days are excluded from the headline comparison. The query behind each figure is listed with this release.

About Harlow Freight. Harlow Freight makes quoting and booking software for business shippers across the US and Canada.

Claim 1 · evidence

q_0412 refrigerated_midwest_yoy

Query

SELECT
  CASE WHEN b.booked_at >= DATE '2025-10-01'
       THEN 'latest_12m' ELSE 'prior_12m' END AS period,
  COUNT(*)                         AS bookings,
  COUNT(DISTINCT b.account_id)     AS accounts
FROM bookings b
JOIN lanes    l ON l.lane_id    = b.lane_id
JOIN accounts a ON a.account_id = b.account_id
WHERE b.freight_tag = 'refrigerated'
  AND l.origin_state IN ('IL','IN','IA','MI','MN','MO','OH','WI')
  AND b.booked_at >= DATE '2024-10-01'
  AND b.booked_at <  DATE '2026-10-01'
  AND a.created_at < DATE '2026-10-19' - INTERVAL '90 days'
GROUP BY 1
HAVING COUNT(DISTINCT b.account_id) >= 50;

Result

period	bookings	accounts
prior_12m	21,400	1,212
latest_12m	72,760	1,840

aggregate · cohort 1,840 accounts · floor 50

Method

Bookings tagged refrigerated, originating in eight Midwest states, latest 12 months compared with the same period last year. Accounts created in the last 90 days are excluded. 72,760 against 21,400 is an increase of 240%.

A claim that cannot show its query cannot go in the draft. There is no setting that relaxes this.

Scope is set by your security team

Nine tables in. Four tables out.

bookingsin boundsbooking_id, account_id, lane_id, freight_tag, booked_at +13 more

shipmentsin boundsshipment_id, booking_id, weight_band, delivered_at +12 more

quotesin boundsquote_id, account_id, quoted_at, accepted_at +8 more

accountsin boundsaccount_id, region, plan, created_at +7 more

lanesin boundslane_id, origin_state, dest_state, mode +5 more

product_eventsin boundsevent_name, account_id, surface, occurred_at +10 more

api_requestsin boundsendpoint, account_id, status, requested_at +6 more

invoicesin boundsinvoice_id, account_id, amount_usd, issued_at +9 more

subscriptionsin boundsaccount_id, status, started_at, ended_at +5 more

customer_contactsexcluded8 fields, not read

support_transcriptsexcluded6 fields, not read

payment_methodsexcluded7 fields, not read

employee_recordsexcluded15 fields, not read

Your team chooses the scope before anything connects, and changes it whenever.

Excluded tables stay visible here so a reviewer can see what we were not allowed to read.

Every figure is an aggregate with a minimum cohort size you set. Ours is 50 accounts by default.

Access is read-only. DataPress cannot write to your warehouse and cannot export row-level data.

Most of what makes this safe is what it cannot do, so we show you that first.

It turned one down this week.

Rejected pitch · not drafted

Proposed angle: Harlow customers are switching away from a named competitor.
Data that prompted it: Accounts that connected a second carrier integration in the last six months increased their Harlow bookings by 46% over the same period. cohort 620 accounts
Rejection note: Declined. The data shows these accounts used Harlow more. It does not show that they used anyone else less, because Harlow cannot see bookings made elsewhere. A switching claim needs data outside your scope, so this pitch will not be drafted.

An engine that finds a story in everything is an engine that will eventually get you a correction.

One number, every month, yours.

Some findings repeat. Turn one into a standing index and it publishes on the same day every month with the same method, which is how a number becomes something journalists check rather than something you announce.

The Harlow Decision Index

1.6 days

Average days from quote to booking, every accepted quote in the month, all accounts. Method locked.

cohort 12,000 accounts · next publication 2 November 2026

How a story becomes a draft

01
The scan
Weekly, inside your scope, looking for patterns against your own history.
02
The pitch
Headline number, chart, method, cohort size, confidence.
03
The angle
You pick the format: press release, post, blog, quarterly report. The evidence does not change, only the framing.
04
The sign-off
A named reviewer approves. Their name goes on the record.

Per connected source, per month.

Priced on sources connected rather than drafts produced, because charging per draft would push us to find more stories than your data actually contains. The scan, the feed, the drafts and the review trail are all included.

no figureSuppressed. Cannabis-adjacent freight, two states.

31 accounts. Below your threshold. Not published.

Look at a week of someone else’s data.

Four stories, one suppressed, one rejected, and a draft with every claim traceable. About two minutes.

See a story feed