A vendor's installed base is the set of customers that currently use its product, counted as accounts or units. It shows how widely the product is used and how large the pool is for renewals, upgrades and add-ons. From outside, it can only be estimated, from visible traces of use rather than the vendor's contracts.
Estimating it from outside
For a web technology the traces are tags, scripts, response headers and DNS records on company websites. Count the websites where the technology is present on a date, and track additions and removals between dates. Net change is additions minus removals, and prevalence is the websites with the technology per 100 websites in a defined cohort. Fix the cohort, as a same-store cohort does: an index of websites that grows by itself makes every count rise without any change in the market.
What the count is not
Websites are not customers. A group can run one tool on many domains, and one account can appear under several. Detection is also partial: a tag that loads only after cookie consent, or only on pages other than the homepage, can be missed, and a tool used behind a login is invisible. Read the figure as web-visible use, and compare it with itself over time more than with the vendor's reported customers.
In Fokals data
Technology Stack delivers each website and technology with a first-seen date, a last-seen date and the date a technology went missing. Technology Changes records a dated event for every adoption and removal, with the first observation of a site set as a baseline, so every event is a real change. Market Series gives weekly added, removed and prevalence series by industry, country and size band, on same-store cohorts. Net change over a window comes from the events:
select
sum(case when change = 'added' then 1 else 0 end) as added,
sum(case when change = 'removed' then 1 else 0 end) as removed
from company_tech_events
where category = 'technology'
and key = :technology
and observed_at >= :window_start
and observed_at < :window_end;Technology Changes also records changes seen in DNS records, so the installed base of a mail or hosting service is covered alongside web tools. The marketing stack dataset holds both datasets, and the guide to measuring a vendor's installed base builds the weekly view.
Related terms
Losses from an installed base are measured as technology churn. Its size feeds a total addressable market, and it is counted from technographic data.
Frequently asked questions
How do you measure a competitor's installed base?
You cannot see their contracts, so measure what is visible. Count the websites where their technology is detected, record additions and removals over time, and express the result per 100 websites in a fixed cohort. Treat the count as web-visible use, not as customers. Where the vendor reports customer numbers itself, compare the two trends rather than the two levels.
What is the difference between installed base and market share?
Installed base counts the customers or units using a product. Market share divides that count by a defined market, such as all companies in an industry and size band, so it needs a denominator. Without a stated denominator, a share cannot be compared between sources, and a share taken from a growing list of websites will drift upward for reasons unrelated to the product.
Is installed base the same as the number of users?
No. Installed base counts the customers or units that have the product, while users are the people who use it. One customer can have ten users or ten thousand. For a technology detected on a company website the unit is the website, so the count says how many sites run the tool and nothing about how many people use it.
The queries and code on this page are examples to adapt. Test them in your own environment before you rely on them.