Baki Bilişim
TR Free Audit

How to choose a corporate digital agency: ask for measurements, not claims

The market is saturated with claims: the same promises, the same logo walls, the same unverifiable numbers. We move the axis to verifiability. Every figure published on this page carries its measurement date, its measurement tool and a verification route you can repeat yourself.

Free Digital Asset Audit Book a Discovery Call

Last updated: 6 proof mechanisms · 12-question selection checklist Measurement date: 2026-07-29

How do you assess an agency with no published case studies?

Answer

We have no published corporate case studies yet; we are taking on our first corporate engagements. That is why we offer measurement rather than references as proof. Assess an agency by the measurable state of its own digital assets: the score, the measurement date, the measurement tool, and a verification route you can repeat without our permission.

We have no published corporate case studies yet; we are taking on our first corporate engagements. That is why we offer measurement rather than references as proof.

We put that sentence at the top of the page, because in a procurement process the most expensive surprise is the one you learn about late.

A reference is one form of proof, not the only one. As a corporate buyer you cannot verify a reference list in practice: you cannot tell which part of the work belonged to the agency and which to the client's internal team, in which year it was delivered, or who has touched the site since. A logo wall is a row of images with a verifiability of zero.

A measurement is the opposite. If you do not believe a score on this page, put the same address into PageSpeed Insights and measure it yourself in about ten seconds. The result does not depend on our consent: it can be disputed, it can be repeated, and it carries a date. Our entire proof architecture is built on that difference.

The second half of honesty is writing down what we do not undertake. We give no ranking guarantee, we claim no control over the output of generative AI engines, and we do not commit to the decisions of third-party platforms. None of those are things we measure or can promise.

Which six proof mechanisms do we publish instead of references?

Answer

Six mechanisms: open methodology, technical depth content, measured live scores, a free technical audit, a transparent working and contract model, and founder and accountability visibility. Each is tied to a page you can check for yourself, and each carries its own stated limit. Proof without a stated limit is marketing, not proof.

  1. Open methodology

    Every stage of our six-stage working model is published in writing: what is done, which deliverable it produces, which criterion measures it, and which party is accountable. You can read the process before buying anything and compare it against your own internal procurement process.

    Where you see it
    How we work — six stages with deliverable lists and measurement criteria
    What it proves
    That scope and acceptance criteria are known before the contract, so surprises are reduced systematically rather than case by case.
    Its limit
    A methodology alone does not guarantee an outcome. That is why every stage ends in a numeric measurement criterion.

  2. Technical depth content

    Our AEO, GEO, Core Web Vitals and Schema.org guides are implementation documents rather than sales copy: they contain code samples, verification steps, sources and access dates. The cheapest way to judge what an agency actually knows is to read what it publishes for free.

    Where you see it
    Knowledge centre · What is AEO · What is GEO · Core Web Vitals · Schema.org · glossary
    What it proves
    That we can apply the work we describe. The steps in the guides are implemented on this site, so you can hold the two side by side and check.
    Its limit
    Content indicates competence; it is not a delivery guarantee. A delivery guarantee comes from the acceptance criteria in the contract.

  3. Measured live scores

    This site's Lighthouse scores, Core Web Vitals figures, page weight, third-party request count and structured data validity are published together with the measurement tool, the device emulation and the measurement date. A row we cannot measure is not published at all.

    Where you see it
    The proof page — with the measurement masthead, the measurement history and step-by-step verification
    How you verify it
    Put the same address into PageSpeed Insights and the Rich Results Test; the result does not pass through us.
    Its limit
    Our own site's score is not a guarantee that the same score will appear on yours. Infrastructure, content volume and integrations differ, so target values are written per project.

  4. Free technical audit

    You test the claim on your own site rather than on ours. The audit report itemises AEO answerability, AI crawler access, Core Web Vitals figures, structured data validity and WCAG 2.1 AA gaps, each with what was checked and how it was measured.

    Where you see it
    Free digital asset audit — scope and delivery time are stated on the page
    What it proves
    Whether we can find the problem. The report is yours to keep; you may take the findings to another agency.
    Its limit
    The audit is free and creates no obligation to engage us. We also audit requests below our budget threshold, then point them to studios of a suitable scale.

  5. Transparent working and contract model

    Scope and deliverable list, response and intervention times, reporting cadence, source code and design copyright, technology handover, exit terms, data protection responsibilities and subcontractor use are written into the contract and explained openly on the site.

    Where you see it
    How we work — the contract and transparency section
    What it proves
    That we do not manufacture dependency: copyright is yours, accounts are in your name, handover terms are known in advance.
    Its limit
    Ranking guarantees, control over generative engine output and third-party platform decisions sit outside what we undertake. The contract says so in the same words.

  6. Founder and accountability visibility

    Who does the work is not hidden. The founder, Erdi Baki, appears by name with his areas of expertise and direct contact details, so you know your counterpart before the first call. We publish no fictitious branch address, no invented team page and no stock images standing in for staff.

    Where you see it
    About · verified address and telephone on the contact page
    What it proves
    That accountability sits with a named person, and that you know who stands behind the work.
    Its limit
    A small team raises a business continuity question. We manage it with documentation, source code handover and written exit terms; the detail is in the questions below.

What the six mechanisms have in common is this: not one of them rests on our own say-so. Each is tied to a page, a file or a contract clause you can check independently.

What are this site's measured scores?

Answer

The figures below were measured on this page's own address under mobile emulation. Every row carries its measurement tool and measurement date, and most carry a direct verification link. No unmeasured row appears here: we publish no estimates, no rounded assertions and no line that says a value is coming soon.

  • Lighthouse Performance (mobile) 100/100 Lighthouse 13.4.1 · mobile emulation · 2026-07-30
  • Lighthouse Accessibility (mobile) 100/100 Lighthouse 13.4.1 · mobile emulation · 2026-07-30
  • Lighthouse Best Practices (mobile) 100/100 Lighthouse 13.4.1 · mobile emulation · 2026-07-30
  • Lighthouse SEO (mobile) 100/100 Lighthouse 13.4.1 · mobile emulation · 2026-07-30
  • Largest Contentful Paint (LCP) 1.5s Lighthouse 13.4.1 · mobile lab · 2026-07-30
  • Cumulative Layout Shift (CLS) 0.00 Lighthouse 13.4.1 · mobile lab · 2026-07-30
  • Total JavaScript weight of the page 4.0KB gzip Build report · 2026-07-30
  • Third-party requests on first load 0 Browser network log · 2026-07-30

These rows come from the same data source as the proof page. The measurement masthead (tool versions, device and network emulation, method), the separation of lab and field data, the date-stamped measurement history and the step-by-step verification instructions all live there.

See every measurement and the method

When your digital work is split between two agencies, which jobs end up with no owner?

Answer

What ends up with no owner is the engineering side of visibility. A marketing agency produces content and advertising; a software agency writes code. Schema.org data, an LCP commitment, hreflang, AI crawler policy, accessibility and performance regression appear in neither contract, so they fall into the gap between the two teams.

The ledger below names no company. It compares two common operating models as categories. It is the full version of the seven-row summary on the home page. In your current set-up, how many of these twelve rows have an owner in writing?

Who does it? · Expanded responsibility ledger · 12 rows · 2026-07-29
Responsibility Marketing agency Software agency Baki Bilisim
Who writes the Schema.org structured data? Partly, usually via a plugin Out of scope Responsibility written into the contract
Who commits to a numeric LCP target? Out of scope Partly, target not expressed as a number Responsibility written into the contract
Who implements hreflang and canonical in a multilingual structure? Partly Partly Responsibility written into the contract
Who sets the AI crawler access policy? Out of scope Out of scope Responsibility written into the contract
Who prepares llms.txt and the machine-readable company summary? Out of scope Out of scope Responsibility written into the contract
Who writes the answer-first content structure and the question inventory? Partly, content is produced but the structural decision is not Out of scope Responsibility written into the contract
Who keeps mobile app content and site content in sync? Out of scope Partly, content ownership undefined Responsibility written into the contract
Who produces the dealer and branch location pages? Partly, template duplication Partly, empty template Responsibility written into the contract
Who audits NAP consistency across site, business profiles and directories? Partly, no regular audit Out of scope Responsibility written into the contract
Who takes on WCAG 2.1 AA accessibility? Out of scope Partly, no acceptance criterion Responsibility written into the contract
Who monitors performance regression after launch? Out of scope Out of scope Responsibility written into the contract
Who hands over the source code, the documentation and the account permissions? Out of scope Partly, handover terms not in writing Responsibility written into the contract

Responsibility written into the contract Case by case, without an acceptance criterion Out of scope

The ledger is a scope map, not an accusation: as long as every row has an owner, it does not matter how many agencies the work is split across. The problem is the rows nobody ever discussed. Our own answer to these twelve rows is written in the scope matrix of eleven services and in our working model.

How to choose a corporate digital agency: the 12 questions to ask

Answer

Choose a digital agency by the verifiability of its answers rather than the quality of its pitch. The twelve questions below are written to be put to any agency, and they include us. A good answer always contains a number, a date, a document or a contract clause; a weak answer is general and cannot be checked.

Copy the list and send it, unchanged, to every agency you invite to tender. Once the answers sit side by side, the difference comes from verifiability rather than presentation quality. Procurement and IT teams drafting a technical specification can use the list as it stands.

CORPORATE DIGITAL AGENCY SELECTION CHECKLIST · 12 QUESTIONS
Source: Baki Bilisim · 2026-07-29
www.bakibilisim.com/en/why-baki-bilisim/

01  What are your own site's mobile Lighthouse scores, and when
    were they measured?
02  Which tool and device emulation produced that measurement, and
    can I repeat it myself?
03  Which numeric acceptance criteria will be written into the
    contract at delivery?
04  Who will own the copyright in the source code, the design files
    and the content?
05  In whose name are the domain, hosting, app store and analytics
    accounts opened?
06  Who writes the Schema.org structured data, and in which
    validator will zero errors be shown?
07  Who decides the AI crawler access policy, and what follows from
    that decision?
08  Who implements hreflang and canonical, and who tests them?
09  Is WCAG 2.1 AA accessibility in scope, and how is it verified?
10  If we part ways, what do you hand over, in what format and
    within how many days?
11  How often do you report, on which metrics, and do you guarantee
    rankings?
12  Who does the work, who is my contact, and do you use
    subcontractors?

  1. What are your own site's mobile Lighthouse scores, and when were they measured?

    What a good answer looks like
    All four scores are given as numbers, with the measurement date and a clear separation between mobile and desktop.
    Warning sign
    Phrases such as “our scores are high” with no figure, a missing measurement date, or the question deflected onto client sites.
    Our own answer
    Four Lighthouse scores with tool and date on the proof page and in the proof panel above.

  2. Which tool and device emulation produced that measurement, and can I repeat it myself?

    What a good answer looks like
    The tool name and version, the device and network emulation and the lab–field distinction are stated, and the measured address is not hidden.
    Warning sign
    Screenshots with the URL cropped out; lab data presented as if it were real user data.
    Our own answer
    The measurement masthead — tool versions, emulation and method — is written on the proof page.

  3. Which numeric acceptance criteria will be written into the contract at delivery?

    What a good answer looks like
    Thresholds are stated: mobile Lighthouse Performance ≥ 95, LCP < 1.8 s, CLS < 0.05, 0 validator errors, WCAG 2.1 AA.
    Warning sign
    Promises made in adjectives that cannot be measured; acceptance criteria absent from the proposal; a “we will look at it after launch” posture.
    Our own answer
    An acceptance criteria table on every service page; delivery conditions in our working model.

  4. Who will own the copyright in the source code, the design files and the content?

    What a good answer looks like
    Copyright passes to the client on payment, and the answer points to the clause number in the contract.
    Warning sign
    Code handover never discussed; dependency on a closed system that runs only inside the agency's own panel and cannot be moved out.
    Our own answer
    Source code and design copyright belong to the client; the condition is written in our working model.

  5. In whose name are the domain, hosting, app store and analytics accounts opened?

    What a good answer looks like
    Every account is opened in the client's name; the agency is an authorised user whose access can be revoked in one step.
    Warning sign
    Accounts opened in the agency's name; access becoming a negotiating position at the end of the relationship.
    Our own answer
    Accounts are opened in your name and we act as an authorised user — including app store account ownership.

  6. Who writes the Schema.org structured data, and in which validator will zero errors be shown?

    What a good answer looks like
    The accountable person is named, the validator is named, the target is zero errors, and regression checking on content updates is explained.
    Warning sign
    The answer “the plugin handles it”; page-type mapping, entity identifiers and graph consistency never mentioned.
    Our own answer
    Graph architecture and validation in the structured data service; the applied version is in this page's source.

  7. Who decides the AI crawler access policy, and what follows from that decision?

    What a good answer looks like
    The decision is stated to belong to the client; the consequences of allowing and of blocking are described without bias; robots.txt accountability is made explicit.
    Warning sign
    No awareness of the subject at all, or an unexamined reflex to “block everything” or “allow everything”.
    Our own answer
    The policy decision and its consequences sit in the GEO service; our own decision is visible in robots.txt and llms.txt.

  8. In a multilingual structure, who implements hreflang and canonical, and who tests them?

    What a good answer looks like
    Reciprocal hreflang tags, the x-default decision, the canonical relationship and the post-launch test method are each described.
    Warning sign
    Reliance on the automatic output of a translation plugin; one-way tags between language versions with no reciprocal counterpart.
    Our own answer
    Multilingual architecture and hreflang setup in the SEO service; the Turkish and English versions of this site are reciprocally tagged.

  9. Is WCAG 2.1 AA accessibility inside the scope, and how is it verified?

    What a good answer looks like
    The checklist and the tool are named; contrast ratio, keyboard navigation, visible focus and alternative text are counted separately.
    Warning sign
    Accessibility described as a layer added afterwards, or as something an overlay widget fixes on the surface.
    Our own answer
    WCAG 2.1 AA is an acceptance criterion in every project; this site's check result is on the proof page.

  10. If we part ways, what do you hand over, in what format and within how many days?

    What a good answer looks like
    What is transferred is listed — code repository, documentation, design source files, account permissions, content backup — and the period is stated in days.
    Warning sign
    The question deflected; an extra fee appearing for the handover; documentation that was never produced in the first place.
    Our own answer
    Exit terms — what is handed over, and within how long — are itemised in our working model.

  11. How often do you report, on which metrics, and do you guarantee rankings?

    What a good answer looks like
    The reporting cadence, the metrics and the data sources are written down. No ranking guarantee is given, and the reason it cannot be given is explained.
    Warning sign
    A position promised for a specific keyword; a report consisting of one traffic chart; metrics left undefined in the contract.
    Our own answer
    We give no ranking guarantee; reporting metrics are itemised in the SEO service and the AEO service.

  12. Who will actually do the work, who is my contact, and do you use subcontractors?

    What a good answer looks like
    The people doing the work are named; subcontractor use is stated plainly, together with the confidentiality obligation that covers it.
    Warning sign
    The team in the sales meeting differing from the team doing the work; a contact who changes frequently; no accountable person ever named.
    Our own answer
    Your counterpart is the founder, Erdi Baki; direct contact details are on the contact page.

You can put all twelve questions to us as well; the answers are already written across this site. The point of the list is not to favour us but to move the comparison onto measurable ground. Definitions of the terms are in the digital visibility glossary.

Why is there no logo wall, testimonial or project counter on this site?

Answer

Because a visitor can verify none of them. A proof image that cannot be verified is decoration rather than proof. We built no component for these elements at all; in their place we put scores stamped with tool and date, acceptance criteria that enter the contract, and a named accountable person.

Common proof imagery · why it is absent here · what stands in its place
Common proof imagery Why it is absent here What stands in its place
A wall of client logos A logo shows neither the scope of the work, nor its year, nor who actually did it. A visitor can verify none of it. Live scores stamped with measurement tool and date (proof)
Testimonials with no name or organisation Praise whose source cannot be checked gives the buyer no new information; it only fills a space on the page. A founder profile with a name and stated accountability (about)
Counters for completed projects and satisfied clients A counter has no definition, no scope and no measurement method; the number is as free to grow as it is to shrink. Measurement rows stamped with date and tool (the S2 component)
Award and badge graphics Without the awarding body, the assessment criteria and the year, a badge is only an image. Verification links that can be repeated with independent tools
Treating elapsed time as competence Years passed do not measure current technical competence; most of the AEO, GEO and Core Web Vitals agenda belongs to the last few years. Technical guides published at implementation level (knowledge centre)
An invented “representative” case study A fabricated example misleads the buyer and collapses at the first technical question; once it collapses, every other sentence becomes suspect. This site itself, as the one example you can verify

These are not a matter of style but a boundary. Manufacturing proof can lift conversion in the short term; in corporate procurement, however, a single unverifiable claim is enough to remove an entire proposal from consideration.

When our first corporate engagements are complete we will publish case studies — with the client's written consent, and with the scope, the date and the before-and-after measurements included. Until that day this section stays exactly as it is.

Questions we are asked about trust and differentiation

Why should I trust you when you have no published case studies?

We propose building trust on verifiability rather than on references. We have no published corporate case studies yet; we are taking on our first corporate engagements and we do not hide that.

In return we publish our own site's Lighthouse scores, Core Web Vitals figures, page weight and structured data validity together with the measurement tool and the measurement date. You can repeat the same measurements in your own browser. You can also test the claim on your own site through the free digital asset audit, without committing to any engagement.

How do I verify your scores independently?

In four steps, all with free tools: enter this page's address into PageSpeed Insights and read the Lighthouse scores on the mobile tab; give the same address to the Rich Results Test and check whether any structured data error appears; run an accessibility check with the WAVE or axe extension; then count the total and third-party requests in the network tab of your browser's developer tools.

The tool versions, the device emulation and the method we used are written in the measurement masthead on the proof page. If your result differs from ours, tell us: we update the value and the measurement date, and we do not delete the previous one.

Why should I work with a single agency?

You do not have to. The point is that when the work is split in two, certain responsibilities end up with no owner. A marketing agency produces content and advertising, a software agency writes code; Schema.org data, an LCP commitment, hreflang setup, AI crawler policy and WCAG accessibility appear in neither contract.

The twelve-row responsibility ledger on this page shows exactly that gap. We are also willing to work alongside your current agency: in that case we separate, in writing, who delivers what and which criterion measures it.

Can a small team handle corporate scale?

Yes, when the scope is chosen correctly. The advantage of a small, focused team is a single chain of accountability: where a decision was taken and who holds the work never becomes unclear. The risk is business continuity, and we do not conceal it.

We manage that risk with four things: every deliverable ships with documentation, source code and design ownership transfer to you, accounts are opened in your name, and exit terms are written into the contract. We decline work beyond our capacity, because accepting a job we cannot deliver is our largest reputational risk.

What exactly separates you from other agencies?

Three concrete differences. The first is scope: build and visibility sit in one contract, so Schema.org data, hreflang, Core Web Vitals thresholds, AI crawler policy and accessibility are named and measured responsibilities.

The second is the form of proof: instead of client logos and testimonials we publish scores stamped with the measurement tool and date. The third is honest scope: we give no ranking guarantee, claim no control over generative engine output, and redirect requests outside our scale to smaller studios.

How do you manage business continuity risk?

By reducing dependency. Copyright in the source code, the design files and the content becomes yours; the domain, hosting, app store and analytics accounts are opened in your name and we are only an authorised user.

Deliverables come with documentation and use widely known standard technologies, so another team can take over. Exit terms are written into the contract: what is handed over, and within how many days, is settled at the start. The detail is in our working model.

Do you work with clients outside Turkey, and how do contracts and invoicing work?

Yes. We deliver in Turkish and English, and the working language of a project is agreed at the outset. Meetings run remotely; the proof, process and contract terms you are reading are the same for every client, wherever they are.

Contracts and invoices are issued from Turkey by Baki Bilisim, and the currency, payment schedule and applicable jurisdiction are agreed before signature rather than after it. For buyers evaluating a Turkish supplier, our address, telephone and the name of the accountable founder are published on the contact page and can be verified.

Further questions are answered on the FAQ page; the approach for your own sector is set out in the solutions section.

See the measured state of your digital assets within five working days.

The audit is free and creates no obligation to work with us. The report itemises AEO answerability, AI crawler access, Core Web Vitals, structured data validity and WCAG 2.1 AA gaps.

Free Digital Asset Audit Book a Discovery Call

We work with corporate-scale, multi-location or multilingual organisations. One-off small jobs fall outside our scope; in that case we point you to smaller studios.