How can we help?
Manifold works with a range of government agencies, market research companies, and partners to source the most reliable and representative data of Canadian consumers. For example,
- Statistics Canada
- Health Canada
- Regional Health Ministries
- Citizenship and Immigration Canada
- Regional School Boards
- BRISC International
- Flyer Distribution Association
- Real Estate Boards/Companies
- Canada Mortgage and Housing Corporation (CMHC)
- Canadian Bankers Association
- Building permit statistics from Municipalities
- Industry Canada
- Online consumer and business directories
- Numeris Canada
- ViviData Canada
- Publication of hospitals, government agencies, and partners
- Open Data Canada, Provinces and Municipalities
- Proprietary survey and research
Occasionally we conduct surveys of cultural communities.
“Population estimates” are different from the “Census population”. They are not directly comparable.
Census population is a count of the total number of people enumerated on the census day. It is often mistaken for the real or actual population. There is an average undercount of about 3% of the population, but this percentage is highly variable and typically higher in rural areas, First Nations, student residential areas and certain ethnic communities. On the Census day, some people were not counted, either because their household did not receive a census questionnaire (for example, if a structurally separate dwelling is not easily identifiable) or because they were not included in the questionnaire completed for the household (for example, the omission of a border or a lodger). Some people may also be missed because they have no usual residence and did not spend census night in any dwelling. Some people did not respond, were traveling or forgot to respond. For example, students away from home, people out of the country at the time, people who do not fill in the survey for any reason, including lack of capacity or lack of official language skills, sick or infirm, part of a group that traditionally does not have a high fill rate (First Nations, temporary workers), etc. Some very small communities report a very high percentage of undercount. A few first nation reserves did not participate in Census 2016, for example, Six Nations, Chippewas and Oneida. The Census 2016 data for municipalities associated with these first nation reserves undercount the population.
Statistics Canada provides a partial explanation of undercount: https://www12.statcan.gc.ca/census-recensement/2011/ref/estima-eng.cfm. Furthermore, in https://www150.statcan.gc.ca/t1/tbl1/en/tv.action?pid=1710000501 Statistics Canada shows that undercounts are adjusted for population estimates at Provincial and Canada levels. On this same page (add data if needed), note the 2nd quarter 2016 population estimate (36,109,487) for Canada, which is 957,759 and more people than the released population count (35,151,728) in the 2016 Census taken that quarter. This is a difference of 2.7%.
Statistics Canada conducts the Reverse Record Check (RRC) after Census to measure census population under-coverage and adjusts population estimates, e.g.,
In “population estimates” we adjust the undercount of Census and estimate the actual population count. The foundation of our estimates is the current and historical Census in the five-year interval. We have developed an enhanced cohort survival model for population projection, which incorporates historical trends/census data, birth and mortality rates, migration and immigration, real estate development and settlement pattern. We also consider students and temporary workers as they are consumers in the communities where they reside. In addition, we consider the tax filer data, Alberta’s annual population count, new postal codes, housing startling statistics from Canadian Mortgage and Housing Corporation, immigration statistics and movers’ data. Our data reflects the current year population estimate is consistent with the estimate from Statistics Canada.
Economic Family and Household are two different measures of population. The total number of Households is much larger than than the total number of economic family. For example, 1-person households are not economic families.
An Economic Family is o a group of two or more persons who live in the same dwelling and are related to each other by blood, marriage, common-law or adoption. A couple may be of the opposite or same-sex.
A Household is a person or a group of persons, who occupy the same private dwelling unit, share household expenditure and do not have a usual place of residence elsewhere in Canada. It may consist of a family group (census family) with or without other non-family persons, of two or more families, sharing a dwelling, of a group of unrelated persons (for example house staff), or of one person living alone.
Our driving time calculator is based on a road network database from Statistics Canada with pre-defined speed limit for highway, major, secondary, local and minor roads. We also consider one-way street and natural barriers like rivers, railways in our calculation.
For any given site, we divide the 360 degrees around the site into segments and calculate the driving distance along with the pre-defined directions within the specified time frame. Thereafter, we connect the endpoints of all directions to build the trade area defined by the driving time.
Our approach was designed to estimate the trade area by driving time. It is not for traveling purposes as traffic data was not considered.
Modeled data is complementary to customer data, transactional data, surveys, mobile and location data. In machine learning, modeled data is part of the deep learning process. At Manifold, we model consumer demographics, spending, lifestyle, product and media usage, shopping behavior and psychographic data to the 6-digit postal code level covering whole Canada. We data-mine survey data from Numeris and Vividata, identify robust and representative patterns in the survey and extrapolate the data to the 6-digit postal code level. Although modeled data may not be as precise as data from survey respondents, it provides estimates of consumer behavior in all 6-digit postal codes and quantifies all Canadian consumers. Most surveys are limited by sample size and are primarily designed to gain a qualitative description of consumers or quantitative estimates of large markets. With an average unit of 15 households, estimates at 6-digit postal code level offer on one side the most granular data available, on the other side, protect consumers’ privacy.
Due to privacy laws, Statistics Canada publishes Census data only at the Dissemination Area level or higher. At Manifold, we build predictive models to interpolate the data down to the 6-digit postal code level. We take consideration of geographic, dwelling structure and neighborhood characteristics of the postal codes in our estimates.
Marketers use modeled data to differentiate 6-digit postal codes, identify best prospects and estimate market potentials, etc.
We provide better and predictive data products developed by renowned mathematicians with thousands of models. Our objective has been always to make predictive data that differentiate consumer behaviour.
We conduct research projects with universities on big data, machine and artificial intelligence for consumer behavioral analysis. Our projects and algorithms have been supported and endorsed by the Natural Sciences and Engineering Research Council of Canada (NSERC). Our data scientists are on editorial boards of scientific journals of Big Data and Analytics.
We have a software development team in house to address clients’ specific requests.
At Manifold, we apply and develop innovative and efficient data mining techniques to create data products. We employ both the traditional statistical methods and the newest data mining technologies to custom solutions for our clients. We have active joint research projects with university professors (Sherbrook, York) on big data analytics. These projects have been supported and endorsed by the Natural Science and Engineering Research Council of Canada (NSERC).
Descriptive Modeling: For hundreds of clients we have provided analytical services on
Market Potential Analysis
Trade Area Analysis
Customer Lifetime Value
Custom Segmentation: We built Customer Behaviour and transactional clusters for fashion retailers: cleo, Rick’s and Bootlegger; Credit cardholder segmentation for Scotiabank, Ally, Duca; and also business cluster of retail stores, e.g., for Sobeys.
Risk Modeling: We have built risk models for Jamaican National Bank and JN General Insurance
- Line of credit;
- Auto Insurance;
- Product Protection Plan;
- Loan applications.
Predictive Modeling: For targeted marketing we have built various types of predictive models.
- Customer acquisition models to predict who will respond to a campaign to buy a product (Credit Card, CAA Membership, Fashion Retailer, Newspaper, Magazine, Cell phone, Satellite TV);
- Up-sell/Cross-Sell models (Fashion Retailers, Saving account of financial institutions, AMITIZA, Audi, Mazda);
- Attrition /Retention models (Rogers Subscribers, CAA members);
- Win-back customers (Not-For-Profits, Canadian Red Cross, Credit Unions).