Data processingPreparation and interpretationRelated hosting of agreed outputsPreston, Lancashire registered

Untidy data in. A clear reading out.

AVC Data Analytics prepares business data so that it can be trusted, processes it into consistent tables and summaries, and then explains in plain English what the figures do and do not show. Agreed outputs can be kept in a controlled hosted location for the people who need them.

What usually arrives
CustomerDateValue
Hartley & Co3/4/25£1,200
HARTLEY AND CO2025-04-031200
Brook Joineryblankn/a
What you decide from
Customer keyDateValue (GBP)
HARTLEY-CO2025-04-031200.00
BROOK-JOINflagged: no dateflagged: no value
1 duplicate merged, 2 gaps reported back to you
Printed tables sorted into colour-tabbed stacks on a pale blue desk
Illustrative image

Problems we are usually asked about

These come up again and again in small and mid-sized British organisations. None of them needs a large programme; all of them need someone to look carefully at the data before anyone presents a number.

  • Two reports, two different totalsFinance and operations each pull a figure from their own export, and nobody can say which is right because the rules for refunds, cancellations or part-months were never written down.
  • The same customer spelt five waysFree-text entry over several years has left a list in which one client appears as separate records, so counts and averages are quietly inflated.
  • Dates that change meaningA file that passed through a US-configured spreadsheet now holds 03/04 as both March and April. Monthly trends built on top of it are not reliable.
  • A monthly report built by handOne person spends two days copying and pasting each month. The process lives in their head and stops when they are on leave.
  • Blank cells with three meaningsEmpty could mean zero, unknown or not applicable. Treating them the same way changes the answer.
  • Outputs emailed as attachmentsVersions multiply, old figures circulate, and no one is certain who has the current file or who should have access.

Services

Four stages, each useful on its own

You can ask for a single stage, for example a one-off clean-up before a board pack, or for the whole sequence as a recurring arrangement. Each stage ends in something you can hold: a file, a document or a link.

Stage A

Data preparation

We profile what you send, agree the rules with you, then clean and reshape the data. Every transformation is written into a change log so the result can be traced back to the source.

  • Matching and merging duplicate records under rules you approve
  • Standardising dates, currencies, codes and category labels
  • Separating genuine gaps from zeros and recording them

Stage B

Processing runs

Once rules are agreed, the work becomes a repeatable job instead of a manual exercise. A monthly or weekly extract goes in; the same checks and transformations run; consistent tables come out.

  • Scripted transformations with input and output counts
  • Exception lists for records that fail a rule
  • Run notes that a colleague can follow

Stage C

Interpretation notes

Numbers rarely speak for themselves. We write a short reading alongside each output: what moved, what is likely noise, which caveats matter and which questions the data cannot answer.

  • Plain-English summaries written for non-specialists
  • Clear statements of coverage, gaps and assumptions
  • Suggested follow-up questions, not forced conclusions

Stage D

Hosted outputs

Where it helps, agreed outputs are made available from a controlled hosted location rather than circulated by email, with access, retention and deletion arranged in advance.

  • One current version, with previous versions labelled
  • Access limited to named people you specify
  • Retention and deletion dates agreed in writing

Read the full service descriptions

What you send, what you receive

Inputs

  • Exports from your existing systems, in CSV or spreadsheet form
  • A short description of the question the figures must answer
  • Any definitions already in use, however informal
  • The name of a person who can confirm business rules
  • Only the fields the work needs: we ask you to remove anything else first

Outputs

  • A prepared dataset in an agreed structure and format
  • A change log listing each rule and how many rows it affected
  • An exception list for records you need to resolve at source
  • Summary tables or a recurring report, as agreed
  • An interpretation note of one or two pages in plain English

Try it

How ready is your data?

A quick, illustrative estimate. Move the sliders to roughly match a file you work with and see which preparation steps would come first. Two further tools, for converting mixed dates and profiling a CSV sample, are on the tools page.

Tool 01 / illustrative

Dataset readiness check

Estimate how much preparation a file needs before anyone should draw conclusions from it. Enter rough figures; no data is uploaded.

A cleaned dataset should change a decision

Preparation is not the goal; it is what makes a decision safe. Three examples of the kind of question prepared data can support, and the caveat that travels with each answer.

Which product lines are really growing?

Once returns and part-months are treated consistently, month-on-month change can be compared fairly. The note will still say if a line is too small for its trend to mean much.

Where are orders delayed?

Joining order and dispatch records on a cleaned key shows where time is lost. If dispatch dates are missing for one depot, we say so rather than guess.

Is the customer list as large as we think?

Matching variant spellings gives a defensible count of distinct accounts. The matching rules are listed so that anyone can challenge or repeat them.

Honest limits

What we will not claim

We are a data processing company, not a regulated adviser. Interpretation notes describe your figures; they are not financial, tax, legal or investment advice.

We do not operate a physical data centre and we do not offer round-the-clock monitoring. Hosted outputs use reputable infrastructure arranged for the engagement, and support runs on UK working hours.

We do not hold industry certifications that we would need to list here, and we will not imply otherwise. If your organisation requires a specific standard, tell us early so we can discuss whether the work fits.

We cannot make poor source data accurate. We can make its weaknesses visible and measurable, which is often the more valuable outcome.

Related hosting

Outputs that stay where they belong

"Hosting" in our registered activity does not mean renting you servers. It means that the prepared datasets, reports and notes we produce for you can live in one controlled place for an agreed period, instead of drifting through inboxes.

Before anything is hosted we agree, in writing, who can see it, how long it is kept, how it is removed and what format you receive if you want it back. When an engagement ends, outputs are returned or deleted on the date agreed and we confirm that it has been done.

How hosted outputs are arranged

Resources

Useful before you call anyone

Our resources page has a glossary of the terms that appear in data work, a checklist for preparing an export safely, and guidance on reading a change log. They are written so that you can use them whether or not you work with us.

Open the resources  /  How to read a result

Support and contact

Tell us about the file, not the platform

The most useful first message describes what the data is, roughly how large it is, where it comes from and what decision depends on it. Please do not attach files or paste records; we will agree a safe route if we need them.

We reply Monday to Friday between 09:00 and 18:00, UK time. Existing clients can raise questions about a run, an output or a hosted file through the support page.

Direct

bi@avcdataanalytics.bid

AVC Data Analytics Ltd
Company number 17408779
Registered office: 131 Williams Lane, Fulwood, Preston, England, PR2 9UF

Use the enquiry form