Aptic
English
Get started
Analysis

Expert-level analysis, run on your own numbers.

Aptic's offices run the analysis a finance lead, a pricing analyst, and a marketing scientist would, on the data you connect. Every figure is calculated the same way every time, kept current as new data arrives, and explained in plain words.

Get started See what it analyzes
How the numbers are made

Calculated, not guessed.

AI is good at explaining and poor at arithmetic. So in Aptic, every figure comes from a tested calculation, and the AI's job is to tell you what it means.

Tested calculations

Each analysis is a fixed, tested calculation. The same inputs always give the same answer, so any number can be checked and repeated.

Run on your data

Nothing runs on data you haven't connected. Each office shows what it's waiting on, and what that data would unlock.

Labeled for what it is

Every figure is marked as observed, calculated, projected, estimated, assumed, or compared with a benchmark, with how sure Aptic is. A forecast never looks like a result.

Says when it can't tell

When the history is too short or the data too thin, the analysis declines to answer and says what it needs, instead of producing a confident guess.

An example

One forecast, and everything behind it.

What a single analysis looks like when Aptic hands it to you.

Illustrative example Projected

Question

Will we have enough cash through the summer?

Cash balance, next 13 weeks

The forecast falls from about $38,500 in week 1 to its lowest point of about $18,400 in week 9, then recovers to about $30,800 by week 13. The low end of the range dips below the $15,000 floor in weeks 8 to 10.

Week
  • Forecast
  • Likely range
  • Your cash floor

Answer: Lowest point in week 9: about $18,400, above your $15,000 floor. The low end of the range dips below it in weeks 8 to 10.

How sure
Medium
Range from
How far off past sales forecasts were

What it means

Two supplier bills and the quarterly tax payment land in the same weeks as the slowest sales. Moving one bill two weeks later, or collecting the two oldest invoices, keeps even the low end above your floor.

Based on

Your bank balance, open invoices and bills from your accounting system, purchase orders, recurring expenses, and the sales forecast.

What would make it surer

Connecting payroll, so pay dates come from the schedule instead of an average.

Checked afterward

As the weeks pass, Aptic scores this forecast against what actually happened, and that record sets how wide the next range is.

What it analyzes

Every part of the business, one set of numbers.

A sample of what Aptic's offices calculate. Each analysis runs only once the data it needs is connected.

Marketing and advertising

Where the money goes, what it returns, and what actually caused the sales.

  • Where ad money is wasted

    Spend with no sales, and clicks that look fine in the ad account but leave the site within seconds.

    Ad clicks matched to site visits

  • What each channel really returns

    Cost per customer and return on spend across paid and organic channels in one view, with how far each figure can be trusted.

    Blended CAC and ROAS

  • Where the next dollar should go

    How returns fall off as a budget grows, and where moving money would earn more.

    Response curves, marketing-mix model

  • Demand you aren't reaching

    Searches with real volume that no ad or page covers yet, and pages one push from the first page of results.

    Search query mining

  • Did the test win?

    A/B results read correctly, even when a test is checked every day.

    Sequential and Bayesian testing

  • Did the campaign cause the sales?

    Market-by-market holdouts, sized before they start and read after they end, to measure real lift.

    Geo holdouts, synthetic control

Sales

What the pipeline will bring in, and which deals you win.

  • Pipeline forecast

    What the open deals will likely bring in, with a range, based on how your own deals have closed.

    Stage calibration, simulation

  • Win rates by segment

    Which kinds of customers you win, with honest error bars on small samples.

    Wilson intervals

  • Time to close

    How long deals take, counting the ones still open instead of leaving them out.

    Survival analysis

  • Revenue concentration

    How much of the business depends on its largest customers.

    Herfindahl index

Customers

Who your best customers are, what they are worth, and who is slipping away.

  • Customer lifetime value

    What each customer is likely to spend over the coming year, and the chance they're still active.

    BG/NBD and Gamma-Gamma models

  • Customers at risk

    Regulars who have probably stopped buying, ranked by what they used to spend.

  • Customer segments

    Best customers, new ones, and the ones going quiet, from purchase history alone.

    RFM segmentation

  • Cohorts and payback

    How each month's new customers keep buying, and how long they take to pay back what they cost to win.

    Cohort analysis

Pricing

What your prices, promotions, and markdowns actually earn.

  • Price sensitivity

    How sales respond to price, measured from your own sales history.

    Price elasticity regression

  • Price test range

    The range worth testing for the most profit, never a single magic number.

  • Promotion check

    Whether a promotion made money once the discount on sales you'd have made anyway is counted.

    Promotion lift

  • Markdowns

    How deep to discount to clear stock by a date, and whether that beats selling it off.

  • Menu and product mix

    Which items are stars, which sell well but earn little, and which to drop.

    Menu engineering

Finance and cash

The work of a finance lead: cash, margin, leaks, and what the business is worth.

  • 13-week cash forecast

    Week by week, from invoices, bills, orders, and expenses, with the lowest week named.

  • Why revenue moved

    Every change split exactly into price, volume, and product mix.

    Variance decomposition

  • Where money leaks

    Duplicate bills, creeping subscriptions, excess discounts and refunds, and sales below cost.

  • Working capital

    How long you wait to get paid, how long stock sits, and how long you take to pay.

    DSO, DIO, DPO

  • Profit by product and channel

    Where the margin is actually made, and where it is given away.

  • What the business is worth

    Normalized earnings, revenue quality, and a valuation range.

Operations and inventory

Stock, staffing, and production, planned from real demand.

  • Reorder points

    Safety stock and reorder points for every item, at the service level you choose.

    Stock policy, EOQ

  • Staffing to demand

    How many people it takes to answer customers within a target wait.

    Erlang C

  • Bottlenecks

    Where work piles up, and which stage slows everything else.

  • Production

    Yield, scrap, equipment effectiveness, and the main causes of downtime.

    OEE, Pareto analysis

  • Materials planning

    Forecast sales turned into the materials and quantities to order.

Product

What people use, what they buy together, and what carries the business.

  • Retention curves

    Whether customers keep using what you sell, month by month.

  • Release impact

    Whether anyone uses what was built, and whether shipping it changed anything.

  • Product classes

    Which products carry the business and which tie up cash.

    ABC analysis

  • Basket affinity

    Which products sell together.

Technology

Security exposures, cloud spend, and room to grow.

  • Security exposures

    Open exposures, graded, with the fixes in order of risk.

  • Cloud spend

    Where hosting money goes, how it is trending, and what can be cut.

  • Reliability and headroom

    How often releases fail, and how much traffic the setup can take before it needs to change.

For your kind of business

Analysis that fits how you make money.

Restaurants and cafes

Menu engineering, labor as a share of sales, margin by item, and promotions that actually paid.

Retail shops

Reorder points, markdown depth, products that sell together, and stock that ties up cash.

Ecommerce

Customer lifetime value, ad clicks matched to site visits, refunds, and promotion lift.

Manufacturing and wholesale

Yield, scrap, equipment effectiveness, supplier costs, and materials planning.

Software and SaaS

Recurring revenue, churn, net revenue retention, payback, and the Rule of 40.

Nonprofits and community organizations

Restricted and unrestricted funds, donor retention, and grant use.

Law firms

Utilization, rate realization, retainer burn, and revenue concentration.

Trades and construction

Crew capacity, jobs stuck in a stage, staffing for call volume, and cash through slow seasons.
All industries
What it won't do

Honest numbers, or none.

  • It won't present a forecast as a result. Projections are labeled and shown with their range.
  • It won't call a pattern a cause. A before-and-after is reported as a before-and-after; cause is claimed only when a comparison group supports it.
  • It won't treat a missing number as zero. A gap stays a gap, and the analysis says what it couldn't see.
  • It won't settle figures your systems disagree on. They are flagged until they reconcile.
  • It won't hide what it skipped. The checks that couldn't run are named, so an empty list never reads as “nothing wrong.”
  • It won't forget how it did. Forecasts are scored when the actual numbers arrive, and that record sets how much weight the next one gets.
  • Readings worth keeping go into Memory as a sentence with its figures, linked to the inputs behind it.
How Memory works
Straight answers

Questions about analysis.

Do I need a data team or a reporting tool?

No. Aptic's offices run the analysis and explain it in plain words. You connect the tools you already use, and Aptic does the rest.

What data does it need?

It starts with what you have. Each office shows what it's waiting on, such as your accounting system, ad accounts, website analytics, or a spreadsheet, and what each one would unlock. Nothing runs on data you haven't connected.

Can the AI make up numbers?

The AI doesn't do the arithmetic. Every figure comes from a fixed, tested calculation over your data, and the AI explains the result. The same inputs always give the same answer.

What if we don't have much history?

Some analyses need a minimum, such as enough repeat customers for lifetime value, or enough weeks where the price changed for a pricing test. Until then, Aptic says what it needs and works from the analyses your data supports, or from the assumptions you state, labeled as assumptions.

How do we know a change worked?

After an approved change, Aptic measures the result. When a comparison group exists, it measures against it. When it doesn't, the result is reported as a before-and-after, with what wasn't ruled out.

Is this a dashboard?

No. A dashboard shows numbers and leaves the work to you. Aptic runs the analysis, explains it, recommends what to do, and, within the limits you set, does the work.

Join the waitlist for Aptic.

We are bringing companies on in small groups. Tell us about your business and we will reach out when there is a place for you.

Get started