Home/Pricing
Indicative rates
What things cost, before you have to ask
Published so you can size a project at two in the morning without talking to anyone, and so a quotation can be checked against a list rather than taken on trust.
They are not quotations. The method drives the cost and the method depends on your question, so the same number of records can be cheap or expensive. Every project is quoted individually against a written scope, and the conversation that produces it is free. All figures in US dollars.
Collection
Fieldwork is priced per completed response and varies with length, difficulty and how hard the population is to reach. Rural and dispersed frames cost more, because of travel time and lower contact rates.
| Item | From | Unit |
|---|---|---|
| Dataset or snapshot | $180 | per deliverable, one-off |
| Completed survey response | $4 – $9 | per response, $600 project minimum |
| Store or outlet visit | $11 | per visit, with photo and GPS |
| Focus group | $320 | per group, moderated |
| In-depth interview | $45 | per interview |
| Instrument translation | $90 | per instrument |
| Transcription and translation | $1.20 | per audio minute, covers both |
| Web and document collection | $0.02 – $0.05 | per row, at volume |
| JavaScript-rendered pages | +60% | surcharge on the per-row rate |
| OCR, printed | $0.11 | per page |
| OCR, handwritten | $0.24 | per page |
Labelling and training data
Priced per item, so a quotation scales with volume rather than guesswork. Double-labelling with adjudication is available where an error carries real consequences.
| Item | From | Unit |
|---|---|---|
| Text classification label | $0.04 | per item, $38 per 1,000 |
| Span or entity label | $0.095 | per item, $95 per 1,000 |
| Bounding box | $0.08 | per box |
| Polygon | $0.26 | per polygon |
| Semantic segmentation | $1.10 | per image |
| Keypoints | $0.14 | per set |
| Audio labelling | $0.90 | per minute |
| Video object tracking | $2.40 | per minute |
| Double-labelling with adjudication | +55% | surcharge |
| Annotation batch minimum | $300 | per batch |
Pipelines, dashboards and models
Built once, then run. The monthly fee covers somebody noticing when a source changes shape, and fixing it before you find out from a broken report.
| Item | Build | Monthly |
|---|---|---|
| Pipeline, single source | $600 | $240 |
| Pipeline, multi-source | $600 | $780 |
| Data platform | quoted | $1,900 |
| Dashboard | $650 | $120 |
| Dashboard, team use | $1,600 | $280 |
| Dashboard, client-facing with logins | $3,200 | $450 |
| Machine learning feasibility study | $900 | — |
| Machine learning build, after feasibility | $3,600 | — |
| Model in production | — | $650 |
Machine learning always begins with the feasibility study, which sometimes concludes that a model is the wrong answer. Saying so is the point of charging for it — the alternative is a build that produces a number nobody should act on.
Research, reporting and time
| Item | From | Unit |
|---|---|---|
| Desk research brief | $650 | per brief |
| Full market study | $2,400 | desk plus primary |
| Market or competitor tracking | $380 | per month, maintained |
| Monitoring | $240 – $950 | per month, by sources and frequency |
| Research report | $450 | per report |
| Full study write-up | $1,200 | report plus analysis |
| Full study with slide deck | $1,900 | report, analysis and deck |
| Statistical analysis | $320 | per dataset |
| Figures and charts | $45 | each, publication quality |
| Proofreading | $14 | per 1,000 words |
| Video report | $320 | short, $140 per finished minute |
| Consulting time | $280 / $42 | per day / per hour |
| Retainer | $780 | per month, three-month minimum |
What a quotation includes
Every one of these, whatever the size of the project. None of them is an extra line on the invoice.
- A written scope and methodology you read and correct before anything starts.
- A sample from your own sources — or, for fieldwork, a pilot wave — before you commit.
- Manual verification of every order before delivery, and back-checks on collected work.
- The methodology note: sampling, response rates, substitutions and limitations.
- One revision round, and deletion or return of source data at the agreed retention date.
- Where a local partner agency is engaged, their name and their cost as its own line.
Nothing here fits?
Custom arrangements are always available — multi-country collection, a dataset nobody has assembled, several disciplines in one engagement, or capacity reserved ahead of a launch. Describe it and we will quote it.